Platform Changelog

​

Highlights

​

Deleted records go to the trash, and can be restored

​

Agents can search the knowledge graph

• rebuilding the entity index from the graph (#3759)
• org-scoped aliases (#3762)
• a reversible merge of duplicate taxonomy nodes (#3767)
• retried entity cleanup when a connector is deleted (#3769)
• the permission check inside the graph query (#3770)
• concurrent enrichment and alias healing (#3771)
• an entity-resolution eval (#3772)
• top-3 merge candidates (#3773)
• further fixes (#3757, #4019) — all @abhishekgml1991

​

AI models: discovery, multi-model add, and prompt caching

​

A locked-down code sandbox and MCP

• Docker access: The app container no longer mounts the Docker socket. A new docker-socket-proxy service holds it and allows only sandbox-image containers with no privileged mode, binds or host namespaces. The app now runs as uid 1000 and keeps its existing paths (#3782, #3811, #3816, #3820). The Helm DinD sandbox goes through the same proxy (#3888 — @abhishekgml1991).
• Network: Sandbox code runs with the network off by default, and when it is on, egress goes through a firewall (#3780).
• Sandbox mode: An unset or unknown SANDBOX_MODE now fails closed (#3725 — @rish231294). local mode requires SANDBOX_ALLOW_LOCAL=true (#3803). Symlink tar members are dropped, and the sandbox user and file-size limit are pinned (#3810). All of these are by @vishwjeet-pawar03 unless noted.
• MCP: Custom STDIO MCP servers run only when the operator sets MCP_ALLOW_CUSTOM_STDIO=true, and catalog commands are pinned (#3887 — @abhishekgml1991).

Connectors remove what the source deleted

• BookStack (#3704)
• Confluence Cloud and Data Center, team and personal (#3705, #3719, #3721, #3823)
• Zammad (#3706)
• Nextcloud (#3707)
• S3-style object stores, including the “Not in” filters (#3708, #3716, #3717)
• Google Drive Workspace (#3709)
• Web (#3712)
• Jira personal (#3934), and Jira on Free plans, which refuse the audit log (#3875, #3905)
• Snowflake (#3688)

​

Indexing: a lane for each connector, and less work per record

​

An All Artifacts gallery, and queued email

Breaking changes & upgrade notes

• Coding sandbox: SANDBOX_MODE must be set. If it is unset or has an unknown value, the sandbox is now unavailable instead of falling back to local (#3725). local also needs SANDBOX_ALLOW_LOCAL=true (#3803). SANDBOX_ALLOW_NETWORK now defaults to false (#3780). Docker Compose installs gain a docker-socket-proxy service under the sandbox profile. The app reaches it through DOCKER_HOST=tcp://docker-socket-proxy:2375 and runs as uid 1000, so custom volume mounts must be writable by that user (#3782). Helm DinD sandboxes go through the proxy too (#3888).
• MCP: existing custom STDIO MCP servers stop running until the operator sets MCP_ALLOW_CUSTOM_STDIO=true (Compose .env or Helm values). STDIO env names on the denylist (PATH, LD_* and others) are refused (#3887).
• Service-to-service auth: the parsing and extraction services require a service token (#3818). The query and connector services no longer send wildcard credentialed CORS (#3890).
• Redis Streams lanes: FAIR_SCHEDULING_LANE_ASSIGNMENT defaults to assigned. Set it to hash on every service to go back to hashed lanes. Events already on a stream are not moved: a connector gets its lane on its first publish after the upgrade, and anything already queued finishes on its old lane. Kafka is not affected (#3982).
• Invites: POST /users/bulk/invite now returns { "queued": true } once the emails are accepted for delivery. A 200 no longer means they were delivered (#2906).
• API status codes and shapes:
  • A member who calls an admin-only route gets 403, not 400 (#3696, #3827).
  • Stats on a connector the caller can’t open return 404, not 403 (#3922).
  • A sync toggle or resync for such a connector returns 404, not 500 (#3955).
  • A resync that can’t be queued returns 503 (#3910).
  • Malformed storage ids return 400, not 500 (#3668).
  • The restore refusal field record_id is now blockedRecordId (#3986).
  • Two knowledge-base record-creation routes on the connectors service, which could only answer 500, are removed (#3933).
• Downloads and secrets:
  • The connectors service’s signed download URLs are bound to the user who redeems them, and still expire after 60 minutes (#3723).
  • On the storage service’s internal download route, expirationTimeInSeconds is capped at 7 days. The default is still 1 hour, and a longer value is refused with 400 (#3790).
  • Connector config responses mask stored secrets and no longer return the owner’s tokens (#3917, #3923). Admins can read secrets back through a dedicated route (#3948).
  • Toolset credentials are hidden from members and masked for admins (#3775).
• Python dependencies: FastAPI moves to 0.136.3, and Starlette is pinned to 1.7.0 (#3817). This changes what callers of the Python service ports see directly. The gateway, UI, Slack and MCP already send the right headers, so they are not affected.
  • JSON requests without Content-Type: application/json get 422 instead of being parsed.
  • URL-encoded forms are limited to 1000 fields and 1 MiB per field, and a form over either limit gets 400. Multipart uploads are not limited.
  • CORS responses echo the request Origin and always send Vary: Origin.

​

Connectors

• Google Drive & Gmail
  • Drive supports external users (#3115 — @R0gue-one), and syncs files that a full listing used to skip (#3678 — @abhishekgml1991).
  • Sharing a file in a filtered folder no longer deletes it (#3873).
  • Gmail web URLs are fixed (#3487 — @R0gue-one).
  • Gmail reads its own OAuth app when it opens a Drive attachment (#3829), and asks only for read-only scopes when it impersonates a Workspace mailbox (#3833).
  • Google toolset tokens refresh with their consent scopes, and the Gmail Drive-download fallbacks are fixed (#3838).
  • Calendar’s consent no longer asks for gmail.send, and Meet’s adds meetings.space.readonly (#3912).
  • A misleading Drive Workspace log line is corrected (#3835).
• Salesforce
  • Sandbox and My Domain orgs sign in on their own login host (#3848), and the setup form suggests a My Domain instance URL (#3830).
  • The Login URL error is now a fixed message written for people (#3904 — @tushar1245, #3906).
  • A file’s stored extension is read from its file record (#3647).
  • Cached search access is cleared when permission edges are rewritten (#4040).
• Confluence — admin date filters apply as set, and the 24-hour look-back runs only from the sync checkpoint (#4042). Folder ancestors are recognised in the placeholder sweep (#3649). “Grant Jira user access” survives toolset registration (#3881).
• Zammad — ticket search pages through /tickets/search, so Zammad 6.4 reads every ticket once (#3807). Every knowledge-base answer is read on Zammad 6.0–6.4 (#3839). Each ticket gets one search window, so a resumed read doesn’t read its ranges again (#3876).
• Linear — a page that times out or is rate-limited is requested again (#3991). A team whose pages never came back is reported, and the deletion checkpoint stays where the scan stopped (#3993).
• Web — the connector recovers from a browser crash (#3815 — @R0gue-one). curl_cffi streaming no longer crashes the connector service (#3755). A page request that never finishes is abandoned and its thread is stopped (#3842). A page keeps its retry when the headless fallback fails (#4004 — @abhishek-pipeshub).
• SMB — the SMB connectors are fixed (#3957 — @R0gue-one), and a renamed file keeps its record (#3645).
• Databases — Postgres tests the credentials before the setup form moves on, and shows PipesHub’s IP (#3985 — @tushar1245). MariaDB skips tables whose content hasn’t changed (#4028). The Redshift driver no longer evaluates int2vector column text (#3895 — @abhishekgml1991).
• Slack — channels sync when the token has no DM scopes, and listing errors fail loudly (#4025). The Slack bot accepts Markdown, text, CSV and TSV attachments (#3683).
• GitHub & GitLab — GitHub rate limits are treated as rate limits, and some quota is left for the user (#4032). GitLab streams items whose titles have characters outside latin-1 (#3938).
• Stored-field fixes after the file-record split
  • SharePoint reads the stored file hash from the file record (#3646).
  • Box reads a shared item’s stored path from the file record (#3648).
  • Nextcloud finds a file’s folder by path (#3650).
  • Local FS reads the file record before retiring one (#3651).
  • Local sync treats the sync root as a literal path (#3898 — @abhishekgml1991).
  • An identical object-store copy gets its own record (#3713).
• Smaller connector fixes
  • RSS saves a feed entry’s new title or link (#3711).
  • Drupal Wiki stops sending back the wiki’s session cookie (#3951 — @abhishekgml1991).
  • Outlook Personal removes deleted mails on ArangoDB (#3855).
  • Airtable and Zoom send client credentials in a Basic header (#3834).
• OAuth registrations
  • Connector and toolset OAuth registrations with the same name are kept apart (#3883, #3902).
  • Connector OAuth apps saved with a toolset’s settings are repaired (#3907).
  • baseUrl is accepted by the OAuth config routes (#3866).
• Connector platform
  • Connector sync is refactored (#3128 — @shrikantNemiwal1).
  • Case-only renames are allowed, and OAuth app authorship is recorded (#3583 — @vishwjeet-pawar03).
  • A connector is marked DELETING before its delete is published (#3840 — @tushar1245).
  • A failed toggle no longer undoes someone else’s newer change (#3915 — @R0gue-one).
  • The crawling manager derives the connector type from the instance (#3658 — @darshan-pipeshub).
  • The graph DB status is reverted when sending a message fails (#3641 — @R0gue-one).
  • Folder path filters allow spaces and commas (#3952 — @abhishekgml1991).
  • A deleted connector’s schedule is removed even when its config can’t be read (#3908).
  • Filter options are capped at 100 (#3910), and /configured and /agents/active page with a true total (#3909).
  • A personal connector is listed only for its creator (#3921).
  • A failed group member removal raises an error instead of returning False (#3937).
  • A connector record and its parent edge are deleted in one write (#3994).
  • Admins are told what disabling sync leaves searchable, and that content stays searchable in exact-text search (#4041).

​

Agents & AI

• Storage Pattern Match — LLM-driven grep over stored records, scoped to what the user can access (#3568 — @vishwjeet-pawar03). Same-named folder and page twins are kept apart (#3761 — @vishwjeet-pawar03). Record JSON is stored as UTF-8, so non-ASCII text matches (#3763 — @abhishekgml1991). Command output and search terms stay out of the info log (#3894 — @abhishekgml1991). A cited grep hit keeps its name, type and quote (#4014 — @abhishek-pipeshub).
• Agent loop
  • A run that reaches its turn limit answers from its last tool results (#3808 — @tushar1245).
  • The agent is told when a withheld tool becomes callable (#4015 — @abhishek-pipeshub).
  • A mistyped record id points back at records already shown, including storage-search results (#4016, #4030).
  • The chat ask-user-question flow is fixed (#3592 — @Pranjal-wq).
• Agent saves
  • A failed save never leaves an agent without its toolsets (#3949 — @tushar1245).
  • A cancelled create rolls back (#4029).
  • An agent is kept when one of its MCP servers can’t be attached (#4049).
• Search
  • Hits from records without a link, such as SQL tables, are kept (#3865).
  • Search falls back to the original query when the AI rewrite fails (#3931).
  • One stalled provider call no longer holds a search for minutes (#3972).
  • Records shared with a role are found on Neo4j (#3710).
  • Cached access is cleared as soon as a connector removes a permission (#4031).
  • A Jira issue key matches exactly (#3914).
• AI model settings
  • Rejected settings get a clear 4xx, the default model is recorded, and follow-up length is limited (#3701).
  • The vector store stays sized for the embedding model in use (#3697).
  • The model health check no longer crashes (#3805), and stops telling admins to uncheck Multimodal over a bad key (#3813 — @abhishekgml1991).
• Tools — Jira get_issue returns the ticket description (#3911 — @darshangodase). Slack tools get permissions (#3722 — @jatingaur18).

​

Fixes

• Knowledge bases & collections
  • A folder delete stays inside the folder (#3679).
  • Duplicate-name checks read the file type from the record (#3802).
  • A move is one write on Neo4j (#3956).
  • A colliding create is retried (#3990).
  • The default knowledge base create converges, and a half-built one is healed (#3996).
  • App browse works on ArangoDB again (#3795).
  • A file’s size shows when it is kept only on its file node (#3801).
  • Long list names show in full (#3724 — @Pranjal-wq).
  • A lock shows when a collection group permission is off (#3676 — @Pranjal-wq).
  • Collection frontend fixes (#3760 — @darshan-pipeshub).
• Users, groups & teams
  • Service accounts are offered in group and team member pickers (#3627).
  • Group names are unique per org, checked on create and rename and enforced in the database (#3638, #3698).
  • A team create converges on one team (#3997).
  • An empty users lookup is skipped, and a bad id gets a plain explanation (#3687).
  • An email edit syncs to the graph (#3436 — @Pranjal-wq).
• Mail — the email-change link now says it lasts 20 minutes, not 14 days (#3692).
• Storage — every caller reads the storage config from the internal route (#3918 — @tushar1245). A missing document on internal routes logs at warn (#4022 — @abhishekgml1991).
• Frontend & desktop — Electron login is fixed (#3602 — @R0gue-one). A notification reloads its conversation even when that conversation is already open (#3946 — @abhishekgml1991).
• Parsing — Windows-1252 text is read before falling back to Latin-1 (#3714).

​

Performance & reliability

• Graph writes are atomic and fail loudly
  • Multi-step Neo4j sync writes are single statements (#3924).
  • Record permission writes are atomic on Neo4j (#3932).
  • Agents, teams and knowledge bases are created all-or-nothing (#4046).
  • A fenced single-field write locks the record first (#3988).
  • ArangoDB write-write conflicts are retried (#3794), and a refused REMOVE fails the delete so it rolls back (#3926).
  • Deleting a collection record removes all its edges (#3874).
  • A mail’s attachments are deleted with it on Neo4j (#3879).
  • Failed reads raise an error instead of returning None (#3672, #3703).
  • A sync point is deleted by its key (#3666).
  • take_back_kept_record_group is defined once (#3920).
  • A Neo4j team KB is found when the list is searched by name (#3925).
  • Service accounts can attach files to chat on ArangoDB again (#3702).
• Indexing status
  • A duplicate takes its twin’s final enrichment state (#3878), and its status is written last so a failed attach is retried (#3594 — @shrikantNemiwal1).
  • A record is marked indexed with an update, not an upsert (#3919 — @shrikant-pipeshub).
  • Only records whose event the queue has moved past are re-published (#3944).
  • Inheritance edges that are removed are counted in the write itself (#4035).
• Services
  • The producer and cache clients are no longer shared across event loops (#3846).
  • A uvicorn worker gets 60 seconds to answer its supervisor (#3939).
  • The query service stays responsive during long streamed tool calls (#3943).

​

Security & authentication

• Advisories — signed download URLs are bound to the user who redeems them (GHSA-cf48-qh3f-g3g9). The sandbox fails closed when SANDBOX_MODE is unset (GHSA-x783-vmqc-7jc9). Ids are encoded where they go into service URLs (GHSA-rfmg-j28j-f635). (#3723, #3725, #3728 — @rish231294)
• Sessions & sign-in
  • The refresh_token grant is bound to the authenticated client (#3639 — @rish664).
  • The SAML session cookie is sent only over HTTPS (#3637), and SAML no longer keeps an in-memory login session (#3700).
  • SAML security fixes (#3765 — @R0gue-one).
  • Ended sessions are refused at the Python services (#3680).
  • Password reset links work only once, even when used twice at the same moment (#3681).
  • A deleted or disabled user’s refresh token gets a plain 401 (#3686).
  • Sign-in attempts are counted before the comparison, and returnTo is validated (#3789 — @R0gue-one).
• Access control
  • Attachment reads are checked for virtual record ids and for history attachments (#3633 — @Pranjal-wq, #3764 — @uncleSlayer).
  • Record updates in knowledge bases are permission-checked (#3640 — @uncleSlayer).
  • stream_record and record delete are scoped to the caller’s org (#3675, #3667 — @tushar1245), and so are connector lists and lookups (#3913).
  • Org-wide schedule routes are admin-only, and queue stats are scoped to the org (#3814 — @darshangodase).
  • Record delete, connector config read and AI model endpoint checks are tightened (#3788 — @shrikantNemiwal1).
  • Domain and “anyone” shares grant no access, and deleted users leave KB sharing lists (#3691).
• Agents — execute_sql_query is limited to allowed connectors (#3677 — @tushar1245). An agent’s sources stay within its configured knowledge (#3751 — @abhishekgml1991). xargs argument injection in Storage Pattern Match is blocked (#3784 — @vishwjeet-pawar03).
• SSRF & untrusted input
  • Internal addresses are blocked when fetching document images (#3632 — @darshangodase).
  • Web and RSS fetches stay off non-public addresses (#3674 — @siddhant-pipeshub).
  • Model endpoints are checked at connect time (#3892 — @abhishekgml1991).
  • Untrusted images are no longer decoded, and empty OAuth scopes are rejected (#4055 — @abhishekgml1991).
  • Regex metacharacters in user and group search are escaped (#3804 — @darshangodase).
  • pdf.js eval is turned off (#3661).
  • Chat asks for a click before it loads images from other sites (#4011 — @Pranjal-wq).
• Slack — the internal Slack config route requires a bot-only token (#3787), and the org is never guessed when hydrating a Slack caller (#3783) — @Pranjal-wq.
• OAuth & secrets
  • The allowed subset of the requested scopes is granted, and usergroup:read is added to MCP_SCOPES (#3731 — @uncleSlayer).
  • Scopes for createConversation and uploadRecords are aligned (#3900 — @siddhant-pipeshub).
  • Shared OAuth app credential trust is fixed (#3768 — @shrikantNemiwal1).
  • Toolset credentials are hidden or masked, and model health checks are admin-only (#3775 — @siddhant-pipeshub).
• Hardening
  • Client-side IP forgery against the rate limiter is closed (#3631 — @uncleSlayer).
  • Server errors return fixed text instead of exception text (#3819 — @shrikantNemiwal1).
  • Nodemailer is upgraded to 10.0.14, and file or URL attachments are refused (#3897 — @abhishekgml1991).
  • onnxruntime telemetry is turned off in every service (#4060 — @abhishekgml1991).
• Dependency patches
  • Node API, frontend build tools and SheetJS (#3662, #3664, #3665).
  • langchain, python-jose and CairoSVG (#3663).
  • FastAPI and Starlette (#3817 — @shrikantNemiwal1).
  • curl-cffi 0.15.0 (#3896 — @abhishekgml1991).

​

Developer experience & infrastructure

• CI & release
  • Every third-party GitHub Action is pinned to a commit SHA (#3685).
  • Integration tests run on pull_request with secrets kept away from forks (#3886 — @abhishekgml1991).
  • Integration-suite guards run as their own required job (#4010, #4036 — @abhishek-pipeshub).
  • CI images are built from GHCR base images (#4045).
  • The sandbox image is published with the app image (#3995).
  • The changelog skill runs under workflow_dispatch (#3659).
  • New F811 redefinitions in tests fail CI (#3852).
  • The PipesHub image version is updated (#3636 — @abhishekgml1991).
• Demo stack — a demo-only stack (#3621), Acme stories for sales, support, marketing, finance and HR (#3585), and an AI judge for demo answers in place of phrase lists (#3669, #3822, #3929, #4000, #4009).
• Integration and end-to-end coverage
  • Auth, sessions and OAuth (#3643, #3671, #3859, #3860, #4008).
  • Permissions and admin routes (#3684, #3689, #3916, #3936).
  • Cleanup after every delete path (#3682, #3756, #3792, #3863, #3869, #3927, #3970, #3971).
  • AI models, agents, MCP and the agent harness on a real model (#3690, #3693, #3797, #3843, #3844, #3845, #3867, #3871).
  • Connector scenario matrix and connector suites (#3644, #3854, #3856, #3857, #3861, #3862, #3864, #3930, #4002).
  • Resilience (#3928, #3989, #4001).
  • Graph edge schemas (#3793).
  • End-to-end UI waits (#3796, #3868, #3870).
  • Stale suites, wait-for-demo-data and route-auth updates (#3673 — @uncleSlayer, #3791, #3809, #3824, #3826, #3954).
  • Connector delete unit tests (#3821 — @R0gue-one).
  • Tests that a later class or function of the same name had replaced now run (#3849, #3850, #3851).
• Docs & cleanup — the README gets the demo video and an install tip (#4033, #4034). Comments and text fragments are cleaned up (#3947 — @darshan-pipeshub, #3958 — @abhishekgml1991).
• Dependencies
  • Dependabot bumps across the Node API, frontend, Python services and integration tests: axios, dompurify, electron, undici, multer, markdown-it, katex, sharp, prosemirror-view, @grpc/grpc-js, engine.io, brace-expansion, source-map-js, @modelcontextprotocol/sdk, sentence-transformers, pyjwt, urllib3, litellm, oauthlib, multidict, pymongo, snowflake-connector-python and fsspec (#3652, #3734, #3735, #3736, #3737, #3738, #3739, #3740, #3741, #3742, #3745, #3746, #3747, #3748, #3749, #3750, #3959, #3960, #3961, #3962, #3963, #3964, #3965, #3967, #3969).
  • @pipeshub-ai/mcp moves to 3.0.1 (#4037).

​

New contributors

• @shrikant-pipeshub made their first contribution in #3919

​

Highlights

​

OAuth apps, personal access tokens and consent require a sign-in session

​

PKCE enforced where the code is issued and redeemed

​

Record convert API: no more writes outside the temp folder

​

Breaking changes & upgrade notes

• OAuth app, personal access token, consent and device-flow routes reject OAuth access tokens and PATs with 403. This covers /oauth-clients/*, /personal-access-tokens/*, POST /oauth2/authorize, /device/verify and /device/consent. Scripts that manage OAuth apps, mint PATs or approve a device login with a token must use a user’s sign-in session instead (#3626).
• PKCE plain is rejected with 400. Public clients must send an S256 code_challenge on both the authorize GET and the consent POST; without it the consent redirects with error=invalid_request and no code. Confidential clients may still omit PKCE (#3630).
• The service-account email domain is reserved. A person can no longer be given an address in service.pipeshub.internal, through invites, user creation or an email change (#3612).
• SDK users: the 1.7.0 SDKs contain breaking changes. See the next section.

SDKs — 1.7.0 (TypeScript, Python, Go)

1.7.0 is a minor version, but it contains breaking changes. Read the list for your language before you upgrade.

​

TypeScript

Renaming a conversation no longer returns messages

// ❌ 1.6.0
// const res = await sdk.conversations.updateConversationTitle({ conversationId, body: { title } });
// const messages = res.conversation.messages;

// ✅ 1.7.0
await sdk.conversations.updateConversationTitle({ conversationId, body: { title } });
const { conversation } = await sdk.conversations.getConversationById({ conversationId });
const messages = conversation?.messages ?? [];

Deleting an agent conversation: conversation can be empty

const res = await sdk.agents.deleteAgentConversationById({ agentKey, conversationId });

// ❌ 1.6.0
// console.log(res.conversation.id);

// ✅ 1.7.0
if (res.conversation && "id" in res.conversation) {
  console.log(res.conversation.id);
}

Some type names changed

// ❌ 1.6.0
// import { Meta, ModelInfo } from "@pipeshub-ai/sdk/models";
// import { IsArchived, GetConversationByIdMessage } from "@pipeshub-ai/sdk/models/operations";

// ✅ 1.7.0
import { MessageFeedbackUpdateResponseMeta, ConversationListItemModelInfo } from "@pipeshub-ai/sdk/models";
import { ListAgentConversationsIsArchived, Message } from "@pipeshub-ai/sdk/models/operations";

Two errors have a new error class

import * as errors from "@pipeshub-ai/sdk/models/errors";

try {
  await sdk.knowledgeBase.uploadRecords(request);
} catch (e) {
  if (e instanceof errors.PipeshubError && e.statusCode === 500) {
    // handle the server error
  } else {
    throw e;
  }
}

Behavior notes (not breaking)

• A user token on 22 endpoints. OAuth apps, personal access tokens, user info, org auth config and reset password now send only the user token. If you set up only an app token, add a user token:
  const sdk = new Pipeshub({ security: { bearerAuth: userToken } });
  
• Stream event data stays a string. Code that calls JSON.parse(event.data) keeps working.

​

Python

Renaming a conversation no longer returns messages

# ❌ 1.6.0
# res = sdk.conversations.update_conversation_title(conversation_id=cid, title=title)
# messages = res.conversation.messages

# ✅ 1.7.0
sdk.conversations.update_conversation_title(conversation_id=cid, title=title)
res = sdk.conversations.get_conversation_by_id(conversation_id=cid)
messages = (res.conversation.messages or []) if res.conversation else []

Deleting an agent conversation: conversation can be empty

from pipeshub_sdk import models

res = sdk.agents.delete_agent_conversation_by_id(agent_key=agent_key, conversation_id=cid)
c = res.conversation
if isinstance(c, models.StoredAgentConversation):
    print(c.id)

Some type names changed

# ❌ 1.6.0
# from pipeshub_sdk.models import Meta, ModelInfo, IsArchived

# ✅ 1.7.0
from pipeshub_sdk.models import (
    MessageFeedbackUpdateResponseMeta,
    ConversationListItemModelInfo,
    ListAgentConversationsIsArchived,
)

Two errors have a new error class

from pipeshub_sdk import errors

try:
    sdk.knowledge_base.upload_records(...)
except errors.PipeshubError as e:
    if e.status_code == 500:
        ...  # handle the server error
    else:
        raise

A citation can have two shapes

from pipeshub_sdk import models

for c in msg.citations or []:
    if isinstance(c, models.CitationReference):
        print(c.citation_id)

Behavior notes (not breaking)

• A user token on 22 endpoints. OAuth apps, personal access tokens, user info, org auth config and reset password now send only the user token. If you set up only an app token, add a user token:
  sdk = Pipeshub(security=models.Security(bearer_auth=user_token))
  
• Conversation streams work. data on ConversationStreamSSEEvent and ConversationMessageStreamSSEEvent is typed Any, as on the other stream models. The SDK decodes each event’s JSON before validation, so the old str type rejected those events.

​

Go

Go 1.26.8 or newer is required

go version            # must print go1.26.8 or newer
go get github.com/pipeshub-ai/pipeshub-sdk-go@v1.7.0

Renaming a conversation no longer returns messages

// ❌ 1.6.0
// res, err := s.Conversations.UpdateConversationTitle(ctx, conversationID, operations.UpdateConversationTitleRequestBody{Title: title})
// messages := res.Object.Conversation.Messages

// ✅ 1.7.0
if _, err := s.Conversations.UpdateConversationTitle(ctx, conversationID, operations.UpdateConversationTitleRequestBody{Title: title}); err != nil {
	return err
}
res, err := s.Conversations.GetConversationByID(ctx, operations.GetConversationByIDRequest{ConversationID: conversationID})
if err != nil {
	return err
}
var messages []operations.Message
if res.Object != nil && res.Object.Conversation != nil {
	messages = res.Object.Conversation.Messages
}

Deleting an agent conversation: Conversation can be empty

res, err := s.Agents.DeleteAgentConversationByID(ctx, agentKey, conversationID)
if err != nil {
	return err
}

// ❌ 1.6.0
// fmt.Println(*res.AgentConversationDeleteResponse.Conversation.ID)

// ✅ 1.7.0
if c := res.AgentConversationDeleteResponse.Conversation; c != nil && c.StoredAgentConversation != nil && c.StoredAgentConversation.ID != nil {
	fmt.Println(*c.StoredAgentConversation.ID)
}

Some type names changed

// ❌ 1.6.0
// var info *components.ModelInfo
// archived := operations.IsArchivedTrue

// ✅ 1.7.0
var info *components.ConversationListItemModelInfo
archived := operations.ListAgentConversationsIsArchivedTrue

Two errors have a new error type

var errResp *apierrors.ErrorResponse
var apiErr *apierrors.APIError
switch {
case errors.As(err, &errResp):
	// error with a JSON body
case errors.As(err, &apiErr) && apiErr.StatusCode == 410:
	// the 410 from Authenticate
}

Success is always true

// ❌ 1.6.0
// if res.DeleteRecordResponseSchema.Success { ... }

// ✅ 1.7.0
if res.DeleteRecordResponseSchema.GetSuccess() { ... }

A citation can have two shapes

// ❌ 1.6.0
// for _, c := range msg.Citations { fmt.Println(*c.CitationID) }

// ✅ 1.7.0
for _, c := range msg.Citations {
	if c.CitationReference != nil && c.CitationReference.CitationID != nil {
		fmt.Println(*c.CitationReference.CitationID)
	}
}

Behavior notes (not breaking)

• A user token on 22 endpoints. OAuth apps, personal access tokens, user info, org auth config and reset password now send only the user token. If you set up only an app token, add a user token:
  s := pipeshub.New(pipeshub.WithSecurity(components.Security{BearerAuth: pipeshub.Pointer(userToken)}))
  
• Stream event data stays a string. Code that reads *ev.Data keeps working.

​

Agents & AI

• Graph enrichment respects permissions. The parents, children and attachments that enrichment adds to search results (Jira, Confluence, Linear) are now permission-checked in one batch, and their number is capped, so names, URLs and summaries of records a user cannot open no longer reach the prompt. A new filter_accessible_record_ids provider method backs it on Neo4j and ArangoDB (#3590 — @darshangodase).
• Agent tools — behaviour tests for the GitHub, Jira and ClickUp tools, and fixes for the bugs they found (#3582); the agent answers from the workspace’s data when the Acme Corp demo is its only knowledge (#3609) — @shekharkadyan.

​

Fixes

• Kafka — the Kafka consumer used by the connectors and query services no longer loses messages after a temporary handler failure. The indexing consumer was not affected (#3515).
• Google — the personal Drive and Gmail connectors keep the fresher OAuth token after the library refreshes it, and save refreshed tokens to settings, which ends a refresh loop (#3571).
• Web — stored ETag and Last-Modified values are read from the file record (#3622).
• Frontend — three wordings the locale pass changed are restored (#3610).

​

Security & authentication

• Service accounts — a person can no longer be given an address in the reserved service-account domain, on any of the four paths that set an email, and the service-account routes get the same rate limit as the token and OAuth app routes (#3612 — @shekharkadyan).
• Toolset OAuth callback — a failed agent access check is answered with JSON instead of a redirect to a caller-supplied address, which the Node API used to follow on the server (#3611 — @shekharkadyan).
• S3 — the bucket name and region are matched literally when a stored file URL is read back, so a bucket such as my.bucket no longer also matches myxbucket (#3608 — @shekharkadyan).

​

Developer experience & infrastructure

• AWS EKS — S3 on EKS gets a SigV4 preload config map in the Helm chart, EKS values and docs updates, and a bucket purge test (#3634 — @abhishekgml1991).
• Tests and CI — a weekly sustained-load test for search and chat that fails on errors and p95 regressions (#3614); a frontend unit-test coverage minimum (#3577); the shared “test” logger is no longer silenced for later tests (#3613); storage tokens are minted for the user the test client acts as (#3623); chat stream mocks really end or stay open (#3624); a missing SMB folder reads as an empty listing in the integration tests (#3620) — all @shekharkadyan.

​

Highlights

​

Projects

​

Identities for automation: service accounts, service tokens and device-code login

​

New connectors and connector options

• SMB 2/3 and SMB1/CIFS network shares — two connectors that share one walker and record mapper, so both write the same file-share graph model (#3533 — @abhishekgml1991).
• Drupal Wiki (KontextWork) — syncs users, groups, spaces, pages and attachments, with per-space permissions and page hierarchy. This first version does not remove pages deleted in the wiki (#3466 — @darshangodase).
• GitHub Teams and GitLab can be configured with a personal access token instead of OAuth (#3325 — @shekharkadyan).
• S3, MinIO, GCS, Azure Blob and Azure Files get a “Folders” sync filter (#3340 — @shekharkadyan).
• Team connectors authenticated as a different account — when the account a connector signs in with has a different email from the PipesHub user who created it, the creator could see none of the connector’s records. Jira and Confluence (Cloud and Data Center), GitHub Teams, GitLab and Linear now link the creator to that account for that connector only (#3313 — @darshangodase).

​

A failed read is no longer a deletion

​

Search: permissions pushed into the vector query, and better keyword ranking

​

Chat: stop generation, working non-streaming endpoints, and custom skills

​

Errors that say what happened

​

Breaking changes & upgrade notes

• The storage service no longer accepts user tokens. /api/v1/document/* (download, buffer, upload, uploadNextVersion, rollBack, directUpload, delete, metadata, isModified) checked only that the document was in the caller’s org, never the caller’s record permissions, so any signed-in employee (or a PAT/OAuth app with kb:read/kb:write) who learned a document id could read, overwrite, roll back or delete files in private knowledge bases and other users’ chat artifacts. Those routes are removed and answer 404; only the service-token /api/v1/document/internal/* routes remain. API clients should use the permission-checked Knowledge Base routes instead, e.g. GET /api/v1/knowledgeBase/stream/record/{recordId}. User-facing links the backend generates (agent artifacts, SQL query CSV exports on local storage) now point at that stream route, and SQL result exports are saved as versioned artifact records owned by the requesting user. The agent tool get_record_download_url now checks the user’s access to the record before minting an S3/Azure signed URL. Chat messages persisted before this change that embed a /api/v1/document/{id}/download link (local storage CSV exports) no longer download (#3593 — @abhishekgml1991).
• Each organization is limited to 5 admins, counting pending admin invites. Promoting, inviting or editing an invite past the cap is refused with a 400. (#3284 — @Pranjal-wq).
• Sign-in no longer reveals whether an email has an account. An unknown email gets the same status and message as a wrong password or code, and a locked account is refused with that same answer (#3529). SAML sent to the normal sign-in endpoint is answered instead of hanging (#3531), and a multi-step sign-in policy that includes SAML is refused on save (#3510). Clients that matched on the old “blocked” or “no such user” answers need to change.
• An auto-blocked account loses its active sessions at once (#3277 — @krishsingh120), and only the account owner can change its email address (#3272).
• Web-search API keys are no longer sent to members in configuration responses (#3532).
• Keyword scoring changes on existing collections. IDF is turned on in place the first time a collection is written to, or when the query service starts; stored data is not rewritten and no re-index is needed. Expect hybrid search rankings to shift (#3488).
• GitLab requires https for OAuth tokens as well as personal access tokens; plain http needs an explicit opt-in (#3362).
• Each GitHub and GitLab connector instance (personal and team) now syncs exactly one repository. The Repository filter is a single choice, and an instance cannot be saved or enabled without one valid repository. A stored configuration with no repository, two or more, or a not_in operator is refused with the reason, so an instance that synced several repositories must be split into one instance per repository (#3251 — @darshangodase).
• install.sh --upgrade now honours --version and PIPESHUB_VERSION (#3175).

​

Connectors

• GitHub and GitLab — personal-access-token setup for GitHub Teams and GitLab (#3325); one repository per connector instance (#3251 — @darshangodase); https required for GitLab OAuth tokens (#3362); four GitLab sync and token fixes found by behaviour tests against a fake GitLab (#3542); the GitHub Teams “all teams” permission is handled conditionally (#3486 — @darshangodase); integration coverage for GitLab and the GitHub personal connector (#3165 — @darshangodase, #3357).
• PostgreSQL and MariaDB — sync no longer loses, strands or leaks tables (#3332); read-only constraints, pooler support and query fixes (#3349); the database group is never deleted as a leftover schema group (#3352); MariaDB incremental sync fixed alongside new integration coverage (#3345); live PostgreSQL coverage (#3180, #3330).
• Object stores (S3, MinIO, GCS, Azure Blob, Azure Files) — the “Folders” sync filter (#3340); no checkpoint past an object that failed to process (#3353); Azure Files re-indexes edited files and removes deleted ones (#3334) and no longer crashes creating placeholder parents in nested folders (#3395 — @cursor[bot]); two Azure bugs fixed while testing S3, Azure and MongoDB replica sets against real servers (#3367); MinIO and object-store test coverage (#3179, #3199, #3341).
• SharePoint, OneDrive and Outlook — SharePoint now deletes records for items deleted at the source (#3347); six OneDrive sync and access bugs fixed (#3539); OneDrive no longer holds the group delta for a member removal already applied (#3564); microsoft-kiota-http is pinned below 1.13 so Graph requests are retried (#3540); SharePoint test wiring (#3186, #3461, #3476).
• Google Drive and Gmail — a Drive timeout error (#3218 — @R0gue-one); folder-filtered Drive syncs no longer silently skip files (#3547); one folder refused with an unrecognised 403 no longer fails every run (#3560); Drive test stability (#3494).
• Jira and Confluence — seven connector bugs previously pinned as expected failures are fixed (#3579); Jira Data Center keeps project roles when the group list is cut off at the picker limit (#3561); behaviour tests for sync, access and recovery (#3506, #3350).
• Box — nine sync and access bugs found by behaviour tests over a fake Box API (#3553); shared files stay in their owner’s drive and deletes are applied (#3557).
• Nextcloud and BookStack — three Nextcloud sync bugs fixed with coverage against servers in the stack (#3354), then the 12 known sync, delete and download bugs (#3573), and file names containing # or ? (#3537).
• Dropbox, BookStack and Salesforce — records are deleted by key in the delete handlers (#3351); a Dropbox assertion error and its logs (#3223 — @R0gue-one).
• Web and RSS — a re-indexed page advances its record version (#3323); one rate-limited site no longer stalls a whole crawl (#3394); seven crawl bugs found by behaviour tests over a fake website (#3554); robots.txt, removal of gone pages and other crawler gaps (#3563); every redirect is checked before it is requested, even when a site refuses HEAD (#3575); a moved file’s old address is removed when its new one is gone (#3604); Web and RSS integration coverage (#3346).
• Local FS — the connector was refactored (#3139 — @R0gue-one); its exclude filter is fixed (#3358); a sync can reset its own records on ArangoDB (#3406); one record it cannot delete no longer ends the sync (#3413).
• Slack — rotating tokens are renewed, a dead one no longer retries forever, and files download with the token in use after a rotation (#3546, #3572, #3580).
• ServiceNow — a set of connector defects fixed (#3097 — @kemalmahmuljin).
• Zammad — covered against a real Zammad (#3415).
• All connectors — every registered connector is held to the BaseConnector contract (#3178); stream-record response errors are handled (#2783 — @shrikantNemiwal1); connector routes are driven over HTTP, with three bugs fixed (#3505); four small bugs fixed on the workspace connector screens (#3545); crawling-manager fixes for one-time schedules, schedule replacement and remove-all, plus team pictures (#3549).

​

Agents & AI

• Agent loop — stale plan and verify checks, the deep-mode hand-off check and a Drive credential mix-up are fixed (#3504); a reply cut off at the output limit is saved as one answer (#3517); replan can see its plan in plan-then-execute mode (#3534).
• Agent tools — behaviour tests for the Teams, ClickUp and GitHub tools fixed 14 bugs (#3497), with further fixes from the Calendar, Slack and Teams tests (#3558) and the SharePoint, Gmail and Lumos tests (#3587); Knowledge Hub name search and SQL tool edge cases, with real tests for the MariaDB, Redshift and Knowledge Hub tools (#3535); list_files searches by name and stays inside the agent’s sources (#3541); an agent’s name search finds files in its knowledge bases (#3570).
• Agent routes and citations — permission and error gaps closed (#3501). User and org context in system prompts is now gated (#3429 — @abhishekgml1991).
• Demo data (Acme Corp) — demo sources are labelled in answers, search results and record lists, and the demo is offered for removal once real data arrives (#3482, #3569); each person can show or hide it (#3512) and an admin can turn it off for everyone or remove it permanently (#3514); the prompt no longer discards demo data (#3500), made-up source links are not sent (#3576), the pricing question is locked for people who cannot see the answer (#3499), and the golden questions are fixed and guarded in agent mode (#3495). Demo persona accounts on the reserved domain get a starting password (#3264).
• Parsing — EPUB support is back, reading books directly and in bounded chunks (#3518, #3525); six parsing bugs fixed by testing the parsers with real files (#3509); HTML tables parse as grids, one row per <tr> (#3507 — @abhishekgml1991); the YAML parser was updated (#3328 — @jatingaur18).
• Embeddings — a null guard on embeddings (#3305 — @Saneajayy) and a retry for Gemini embeddings (#3519 — @kemalmahmuljin).
• Evals — the live harness no longer reports a run that measured nothing as passing (#3187); the no-sources prompt does not name a tool the model cannot see (#3303); golden agent evals run on pull requests (#3177) and answer-quality evals nightly and weekly (#3412).
• Telemetry — activation events for tokens, MCP clients and answers with sources (#3267); pat_created reports the lifetime the token was minted with (#3589).

​

Fixes

• Chat — a failed chat keeps its real error across a reload (#3296 — @Pranjal-wq); seven bugs found by covering the chat controller (#3496) and six more from the chat, message box, message list, knowledge-base page and connector filter tests (#3503); attachments work in shared chats (#3465 — @Pranjal-wq), and Shared Chats shows who shared each one (#3294 — @Pranjal-wq).
• Frontend — long upload filenames scroll inside the file tree (#3309 — @Pranjal-wq); the collapsed sidebar no longer clips the workspace menu (#3392 — @vansh0211); the knowledge-base sidebar uses one loader for folder lists and clears it on sign-out (#3530); the auth store is imported from config.ts (#3574 — @R0gue-one); bugs found by new unit coverage of chat streaming, sign-in, connectors, Collections and the agent builder (#3339, #3498, #3548).
• Localization — German locale coverage completed (#3428 — @nikdav), locale coverage finished and checked for every language (#3588 — @tushar1245), and strings and UI that the locale pass dropped are restored (#3578 — @abhishekgml1991, #3586).
• Knowledge bases — teams are checked before a knowledge base is shared, and All-team members are not re-added (#3371); non-English file names stay intact on upload (#3370); two upload bugs fixed by covering the knowledge base API over HTTP (#3538); hidden collections are excluded from knowledge hub search seeds (#3337 — @shrikantNemiwal1).
• Users and orgs — concurrent signups on a fresh install no longer collide on a slug and return a 500 (#3321); the user is saved before the user-created event is published, and duplicate emails are refused cleanly (#3281).
• Search — Neo4j search filters no longer match labels nothing stores (#3372).
• Timestamps — naive ISO timestamps are treated as UTC in parse_timestamp (#3470 — @pamod-madubashana).
• Nightly findings — three batches of fixes for failures the nightly and end-to-end runs found (#3439, #3440, #3441).

​

Performance & reliability

• Indexing consumers no longer fail to start or hang on shutdown (#3527), and message consumers skip unreadable messages instead of retrying them, tested against real broker behaviour (#3508).
• ArangoDB keeps one HTTP session per event loop (#3492 — @kemalmahmuljin). The raise_on_error argument is keyword-only on every graph lookup that has it (#3483).
• Configuration store — four bugs fixed in the encrypted etcd store (#3550); watchers receive decrypted values and directory listings stay inside the directory (#3551); etcd login, timeout units, port default, watch shape and create-if-absent outages (#3559).
• Pluggable backends — every backend is tested against a real server in CI, with three bugs fixed (#3374).
• Kubernetes — a guided AWS EKS deployment with chart-owned MongoDB and Redis; the same PR fixes streamed chat answers failing to save on replica-set installs (#3544 — @abhishekgml1991). The Helm chart installs healthy again and is tested in CI (#3373).
• Resilience and load tests — fault injection for connectors and dependency outages (#3369); database restarts, AI provider outage and unwritable storage (#3387, #3457, #3462); deletes that leave data behind and blocks that leave sessions open (#3235); weekly indexing, search and chat benchmarks with baseline comparison (#3377, #3398) and scale and stress runs (#3402).

​

Security & authentication

• Five security fixes found by new sign-in, user and config tests: sign-in now enforces the methods the organisation allows, a later sign-in step can no longer be completed as a different account, a one-time code works only once, the internal settings-reload endpoints no longer return server secrets, and unreadable Slack bot settings are no longer treated as “no bots” and overwritten (#3502).
• Microsoft sign-in checks that an ID token was issued for this app and tenant (#3393). The Azure AD provider config is published under the key the frontend reads (#3297 — @rbales79), and the first SSO login of a JIT-provisioned user no longer fails on the org lookup (#3300 — @rbales79).
• CI keeps the PR-validation checkout token out of reach of pull-request code (#3316).
• Permission tests check what an ordinary member can see with realistic files (#3365), what each person can open in Drive and GitHub (#3364), and detect the account lock without relying on the old “blocked” answer (#3566).

​

Developer experience & infrastructure

• One command for the offline suites — scripts/verify.sh (#3196), with CI now running the Omnigent helper tests (#3188), the frontend unit tests (#3190), the Python integration tests that need no services (#3189) and the Electron local-sync tests (#3191).
• Release automation — release verification, rollback and post-release probing (#3176); the eval install and ArangoDB are checked in the install and release tests (#3368); the arm64 image is built and started from main every night (#3366).
• Integration tests — sharded into parallel jobs (#3326, #3400), run on Neo4j by default on pull requests (#3431 — @siddhant-ota), rotated onto Kafka two nights a week (#3385), and run on Azure OpenAI (#3433, #3437 — @siddhant-ota). They now cover MCP (#3289 — @siddhant-ota), the published MCP package against a real PipesHub (#3450), the pipeshub-ai/examples repository and its Go SDK starter (#3600, #3605), and the SMB suite against a Samba share (#3598). The nightly fails when a connector’s credentials or the storage secret are missing (#3409, #3607), and placeholder connectors are left out of the coverage count (#3356).
• Test reliability and repairs — OAuth app creation retries (#3193); the connector deletion test can fail (#3200); a missing Tuple import (#3315); chat-streaming and stop-generation tests (#3317); user-activity tests (#3320, #3327 — @abhishekgml1991); MinIO images, first from quay.io and then from Chainguard (#3318, #3493); the test client is granted every backend scope (#3342); storage suites (#3410, #3434); admin-only AI model checks (#3418); container-filter tests on all three vector stores (#3432); the graph-entity guard step (#3451); and smaller fixes (#3383, #3384, #3396, #3397, #3414, #3454, #3456, #3460, #3523, #3536, #3552, #3565, #3567, #3555, #3599).
• Browser tests — end-to-end tests fail instead of silently passing, with a fast smoke set (#3359); real browser tests for connecting a data source and setting up an AI model (#3360), and for a cited answer, invites, sharing and agents (#3375); Projects tests (#3343, #3442); and smaller fixes (#3336, #3403, #3404, #3405, #3408, #3475). Retrieval tests search the corpus’s own knowledge base (#3335).
• Coverage — mocha worker coverage is no longer lost at shutdown (#3520); lines that mention “password” are no longer hidden and the Python minimum is 89.8 (#3526); the nightly can run the demo suite on its own (#3522).
• API docs — SSE event data is typed as a JSON object, not a string (#3255); the error envelope carries the request id it already sends (#3430); the storage service-to-service routes are described (#3411); populated citations on create and add-message are documented, and the nightly installs the SMB client libraries (#3603).
• Docs — the README leads with what you can build (#3253) and its Quickstart and FAQ no longer describe two different deployment methods (#3322 — @aBiR1D); CONTRIBUTING gains a labeled-issue path and issue templates (#3258).
• Dependencies — @pipeshub-ai/mcp 2.4.2 (#3311 — @siddhant-ota); anyio and soupsieve in the integration tests (#3348, #3324 — @dependabot[bot]).

​

New contributors

• @kemalmahmuljin made their first contribution in #3097
• @krishsingh120 made their first contribution in #3277
• @aBiR1D made their first contribution in #3322
• @cursor[bot] made their first contribution in #3395
• @vansh0211 made their first contribution in #3392
• @pamod-madubashana made their first contribution in #3470
• @nikdav made their first contribution in #3428

​

Highlights

​

Fair scheduling: every connector gets its own lane into indexing

​

An indexing pipeline that sizes itself to what downstream can take

​

No more records stuck in QUEUED

​

Redis: one connection provider, ready for Redis Cluster and AWS MemoryDB

​

Vector collection strategy and per-org deduplication

​

Chat messages get their own collections

​

S3 with an instance IAM role

​

Breaking changes & upgrade notes

• Redis TLS is now enforced wherever a Redis client is created. Before, only part of the code read REDIS_TLS. A deployment running REDIS_TLS=true against a plaintext Redis will now fail to connect; check that your endpoint supports TLS, or unset the variable. REDIS_TLS_ENABLED and REDIS_TLS_REJECT_UNAUTHORIZED accept true/1/yes/on and false/0/no/off, and any other value fails startup. REDIS_TLS_REJECT_UNAUTHORIZED=false is refused when Redis credentials are set; use REDIS_TLS_CA_PATH instead, which is now read as a file and fails startup if it cannot be read. REDIS_TLS_ENABLED is OR-ed with the TLS flag stored in the admin UI for most clients, but the Node.js Redis Streams and key-value store clients read only REDIS_TLS_ENABLED: if you turned TLS on only in the admin UI, also set REDIS_TLS_ENABLED=true (#3155).
• REDIS_DB is deprecated. It still works with REDIS_MODE=standalone but fails startup in any other mode; use REDIS_KEY_NAMESPACE for isolation. REDIS_KEY_NAMESPACE together with MESSAGE_BROKER=redis is rejected at startup, because it does not cover Redis Streams names (#3155).
• The crawling queue moves from the bull key prefix to {crawling}. The first boot after upgrade removes the old keys and re-creates every connector sync schedule from its stored configuration. If the connector service is not reachable within the startup retry window, the schedules are not re-created until the Node.js service restarts: check that your scheduled syncs are back after the upgrade, and restart the Node.js service if they are not (#3155).
• VECTOR_DB_TYPE=redis is gated on VECTOR_REDIS_MODE (which defaults to REDIS_MODE), so the app can run on MemoryDB while the vector index stays on a standalone Redis (#3155).
• New VECTOR_COLLECTION_STRATEGY setting (default single, the existing behaviour). It is read only on the first startup that finds nothing stored; after that the stored value wins, and a value that contradicts it fails startup. Changing it on a live deployment is an embedding rebuild, not a config edit (#3112).
• MD5 deduplication is scoped to one organization. Content that was deduplicated across orgs re-indexes once per org after the upgrade, so expect a one-off rise in indexing volume and vector-store size. Existing records are not rewritten (#3112).
• Connector deletion is “an administrator, or whoever created it”, for team and personal connectors alike, and connector lookups are scoped to the caller’s organization (#3112).
• The stranded-record sweep is on by default (STRANDED_RECORD_REPUBLISH_AFTER_SECONDS, 3600s); it was off (#3163).
• New and changed indexing settings. MAX_CONCURRENT_INDEXING still overrides the new index budget, and MAX_CONCURRENT_PARSING / MAX_CONCURRENT_INDEXING should normally stay unset. NEO4J_MAX_CONNECTION_POOL_SIZE (default 100) is new. Real Neo4j transactions are available behind NEO4J_EXPLICIT_TRANSACTIONS=true, off by default: try it on staging first and watch the indexing logs for DeadlockDetected. The parsing client’s read timeout now follows RECORD_PROCESSING_TIMEOUT (#3170).
• A document the parsing service cannot parse returns 422, not 500, so it fails as a document error on the first attempt instead of tripping the parsing circuit breaker (#3170).
• Chat storage migration. The first boot copies every conversation and agent conversation into chatSessions and chatSessionMessages. The copy is resumable per document and per collection, and the legacy conversations and agentconversations collections are kept, no longer written to. A legacy document without an orgId stops the migration instead of being copied without one (#3148).
• Python services accept a token signed with the scoped secret only as a service token with the connector:signedUrl, record:content, conversation:create or fetch:config scope. Refresh, password-reset, email-verification, mail, storage and user-lookup tokens now get 401 there. Refresh, password-reset and verification tokens are signed with a key derived from the scoped secret that the Python services never hold (#3283).
• Query-service tuning variables now reach the service. QUERY_UVICORN_WORKERS (new, default 1), PIPESHUB_AGENT_TRANSPORT, PIPESHUB_ACCESSIBLE_RECORDS_CACHE, PIPESHUB_ACCESSIBLE_RECORDS_CACHE_TTL and PIPESHUB_SIGNED_URL_CACHE_SECONDS were ignored by Compose and Helm before. Every default keeps the current behaviour (#3113 — @shrikantNemiwal1).

​

Connectors

• Google Drive and Gmail — blocking Google API calls move onto a shared thread pool (#2886); items outside the workspace are fetched (#3052); streaming a record from a shared drive works in the personal Drive connector (#3160) — all @R0gue-one.
• Jira — the Jira connector was refactored (#2775 — @darshangodase). Jira Data Center and Linear send notifications when a sync, a project or team, a permission read or an attachment sync fails (#3123 — @darshangodase).
• Confluence — “Grant Jira user access” defaults to Yes (#3137 — @tushar1245).
• GitLab — permission fixes for confidential work items (#3248 — @darshangodase).
• OneDrive — blocked users are skipped during sync (#3164 — @R0gue-one).
• Notion — an inherited OAuth app is handled, with better logging (#3222), and nested data sources are handled (#3252) — @shrikantNemiwal1.
• Slack — the connector icon is served from the path it declares (#3107).
• All connectors — operating-system metadata files are filtered out before indexing (#3147 — @R0gue-one); the connector factory can unregister a connector (#3162 — @vishwjeet-pawar03); the connector stats endpoint is fixed (#3133 — @vishwjeet-pawar03).

​

Agents & AI

• Coding sandbox — the sandbox that runs agent code was restructured into Docker, settings, registry and governor modules (#3149 — @abhishekgml1991).
• Embeddings — embedding calls get a timeout and a shared retry policy (#3161 — @abhishekgml1991), and chunks are sized by tokens against the embedding model’s limit instead of by characters, so a long block no longer fails the whole record (#3286 — @rbales79).
• Agent builder — a correct default reasoning effort, with its locale keys (#3166 — @vishwjeet-pawar03, #3173 — @Pranjal-wq); an agent with no knowledge source attached says so (#3269 — @rbales79).
• MCP — fallback MCP configurations are fixed (#3221 — @vishwjeet-pawar03).
• Retrieval — retrieved images are kept after transport deduplication (#3244 — @TatshSiow).

​

Fixes

• Users and workspaces — when auto-join is off, people can ask to join a workspace, with notify and decision emails (#3130, #3158 — @Pranjal-wq); Groups move out of People, with permission placeholders for the open-source edition (#3068 — @Pranjal-wq); inactive users are left out of team member lists (#3171 — @darshan-pipeshub); the invite button is disabled when SMTP is not configured (#3249 — @tushar1245); org signup answers 400, not 500, for validation errors (#3098 — @dineshyadav03).
• Knowledge base — fetching a deleted record no longer answers 500 (#3246 — @Saneajayy).
• Configuration — AI model and storage settings fixes (#3145) and a rework of how stored secrets are redacted in settings responses (#3091) — @vanshg685; the vector-store actions in Labs show under the right condition (#3127 — @vishwjeet-pawar03).
• Graph databases — the ArangoDB app schema and an optional Knowledge Hub search parameter (#3119 — @shrikantNemiwal1); agent-template soft delete on ArangoDB passes the right arguments (#3125 — @Anai-Guo).
• Logging — the query service logs unhandled exceptions in its auth middleware instead of swallowing them (#3224 — @L4XB); a Slack request-verification log line is removed (#3217 — @tushar1245).
• Smaller fixes in the token manager, the installer, the graph providers and the environment templates (#3288 — @abhishekgml1991).

​

Security & authentication

• Service tokens and URL fetches — the Python services stop accepting any scoped-secret JWT as a user credential, and the Skills URL import can no longer fetch arbitrary server-side URLs (#3283 — @abhishekgml1991).
• Cross-org isolation — another org can no longer chat as an org’s service agent or probe whether one exists, signed-URL minting and download check identity, record org and permissions, and local storage rejects any path that leaves the org’s folder (#3118 — @Pranjal-wq). Further org-id fixes in sign-in, mail and configuration (#3109 — @vanshg685), and notification lookups filter by org (#3216 — @R0gue-one).
• Sessions and sign-in — a role change ends the user’s sessions in every workspace that uses the same email (#3169 — @Pranjal-wq); the one-time code logic is hardened (#3102 — @rish664); the OAuth client secret is no longer returned when JIT provisioning is disabled (#3099).
• Toolsets — toolset routes no longer fall back to client-supplied headers for the user and org identity (#3103 — @igennova).
• Reporting — the address for security vulnerability reports changed (#3276 — @rish664).

​

Developer experience & infrastructure

• Installer and Compose — download progress no longer repeats in the terminal (#3086); a MongoDB segfault no longer fails the install or risks data (#3100); the manual quick-start produces a working stack (#3105); a first-run script creates the organization and writes a personal access token to a file without printing it (#3182).
• Docs — the README states what PipesHub is once, at the top (#3168), puts the curl install command first (#3089), drops Bitbucket from the code-search claim (#3088), and moves file-format support and the tech stack to the FAQ and docs (#3256, #3257); the contributor setup guide, the manual Compose examples and the connector playbook are corrected (#3104, #3106, #3111); the Spanish deployment FAQ is updated (#3273 — @defawan); the 0.7.0 changelog was added (#3142 — @rish664).
• Integration and end-to-end tests — Groups Playwright specs (#3087 — @Pranjal-wq); test fixes and updates (#3090 — @jatingaur18, #3101 — @abhishekgml1991, #3156 — @vanshg685, #3172 — @siddhantota); org response validation (#3129 — @R0gue-one); GitHub, GitLab and GitHub Teams environments (#3157, #3146 — @darshangodase).
• Dependencies — Node.js API: nodemailer 9.1.1, sharp 0.35.4, morgan, js-yaml, hono, multer and @xmldom/xmldom (#3209, #3227, #3211, #3228, #3229, #3210, #3205); frontend: sharp and next, js-yaml, vitest, browserslist, fast-uri, @xmldom/xmldom and @humanfs/node (#3213, #3212, #3207, #3132, #3150, #3152, #3151); Python and integration tests: nltk, thrift and snowflake-connector-python (#3135, #3134, #3131, #3124) — all @dependabot[bot].

​

New contributors

• @blockgroot made their first contribution in #3037
• @dineshyadav03 made their first contribution in #3098
• @Anai-Guo made their first contribution in #3125
• @darshan-pipeshub made their first contribution in #3171
• @L4XB made their first contribution in #3224
• @rbales79 made their first contribution in #3269
• @defawan made their first contribution in #3273

​

Highlights

​

Personal access tokens for MCP and API access

​

MCP server on the new agent loop, gated from Labs

​

GitHub teams connector, HackerNews datasource, and a code parser

​

Multimodal embeddings, EPUB, AWS Bedrock, and pasted text

​

Query service throughput, phase two

​

Admin role moves to the user record

​

One-command Docker installer

​

Stateless Docling processing and adaptive parsing

​

Breaking changes & upgrade notes

• Admin role migration. On first boot, admin user-group members are promoted to role=admin and the admin groups are soft-deleted; user JWTs without a valid role are rejected, so users re-login once. Admin checks now read the role from the database (#2939).
• Org-ID backfill for storage records. A startup migration backfills orgId on storage documents in MongoDB (#3059 — @vishwjeet-pawar03).
• Compose container names change. Existing Docker Compose installs recreate every container on the next ./install.sh; names become {project}-{service}-1 (default pipeshub-ai-pipeshub-ai-1), so docker exec pipeshub-ai must become docker compose -p pipeshub-ai exec -T pipeshub-ai …. Data volumes are unchanged (#2634).
• Feature flags move. ENABLE_MCP is now controlled only from Labs (the PIPESHUB_ENABLE_MCP env override is gone); code execution is env-only via PIPESHUB_ENABLE_CODE_EXECUTION and its Labs toggle was removed (#3044). Actions/toolsets are gated by a new ENABLE_ACTIONS platform flag, default on (#3045).
• New query-path defaults. The accessible-records cache is enabled by default (PIPESHUB_ACCESSIBLE_RECORDS_CACHE); direct LLM transports (PIPESHUB_AGENT_TRANSPORT=direct) and signed-URL caching (PIPESHUB_SIGNED_URL_CACHE_SECONDS) are opt-in (#2963).
• Docling API surface. Legacy PDF-processing and block-creation endpoints were removed from the Docling service (#2950).
• Embedding cleanups are now behind a feature flag (#3077 — @vishwjeet-pawar03).

​

Connectors

• GitHub — new teams connector; personal connector refactored onto shared methods (#2989 — @darshangodase).
• HackerNews — new API client and datasource (#3060 — @igennova).
• Jira Data Center / Confluence Data Center — placeholder handling and custom hierarchy support (#2878 — @darshangodase); spacing fix in action names (#3034).
• Linear — placeholder handling (#2880 — @darshangodase).
• Notion — reindexing fix and API resilience (#3022 — @shrikantNemiwal1); placeholder handling (#3085).
• Google Drive / Gmail — record streaming now iterates over all permissioned users instead of only the first (#2965 — @R0gue-one); Gmail shared-mailbox search links no longer always point at the synced owner’s account (#3011 — @aesp2008).
• GitLab — connector fixes (#3058 — @darshangodase).
• Web — S3 storage for the web connector fixed (#2983 — @kaushalwaghela-art).
• RSS — record preview no longer fails with “File URL not available” (#3079 — @uncleSlayer).
• Mail — conversation index on mail records (#2972 — @shrikantNemiwal1).
• A broad sync of accumulated changes across the Node.js API — Slack bot integration, configuration manager, mail templates, storage, and user management (#3028 — @vishwjeet-pawar03).

​

Agents & AI

• Structured JSON output is recovered from validation errors instead of failing the turn (#2948 — @TatshSiow).
• A prompt-injection guard was added to the sub-agent execution rules (#2987 — @abdul-5844).
• Toolset actions for the assistant agent were fixed (#2977 — @vishwjeet-pawar03); dedicated prompts for internal search and web search were removed in favour of the agent loop’s defaults (#3031).
• Reindexing embeddings was fixed (#3080 — @jatingaur18); Qdrant gains off-disk storage support (#2995 — @abhishekgml1991).
• The webSearch schema description is now endpoint-neutral (#2968 — @uncleSlayer).

​

Fixes

• Chat mode selection fixed (#2966 — @vishwjeet-pawar03); streamed markdown tables and citations no longer flicker (#3082 — @tushar1245).
• The skills section was removed and the agent sidebar can now expand (#2978); model names display correctly in the UI (#3063); the search UI option was fixed (#3067 — @abhishekgml1991).
• A red dot now flags when default filters have been changed (#2974 — @darshangodase).
• Record preview skips streamRecord when the record is not preview-renderable (#2979 — @Pranjal-wq).
• Korean and Chinese (Taiwan and Singapore) UI languages added (#3051 — @rish664).
• A users-controller route had its HTTP method type corrected (#2986 — @vanshg685).
• A batch of OSS bug fixes (#3072 — @jatingaur18).

​

Performance & reliability

• Vector cleanup is no longer lost on record delete when the broker publish fails (#3013 — @aesp2008).
• Telemetry pushes no longer hit HTTP 413 (#3073 — @rish664).
• Noisy logs were dropped to debug level across services (#3066, #3070 — @abhishekgml1991), with tests updated to match (#3071 — @igennova).

​

Security & authentication

• An x-is-admin header injection path was closed; admin status now comes from the JWT role (#2985 — @Pranjal-wq).
• The uuid package was replaced with crypto.randomUUID() in the Node.js backend (#3020 — @rish664).
• Dependency bumps: cryptography 48 → 50, authlib, urllib3, idna, setuptools, langchain-classic in integration tests (#2898, #2268, #2253, #2346, #2755, #2271); ip-address / express-rate-limit, undici, fast-uri, engine.io, body-parser in the Node.js API (#2953, #2954, #2957, #2804, #2746); nanoid, tar / electron-builder / app-builder-lib / builder-util-runtime, nodemailer / monocart-reporter in the frontend (#2934, #2806, #2805, #2803, #2730).

​

Developer experience & infrastructure

• AGENTS.md documents the repo for coding agents (#2984 — @shekharkadyan); Windows npm setup clarified (#2998 — @mdakram2002).
• A unified system health-check script (#3048 — @Saneajayy) and a changelog CI job (#3061 — @rish664).
• OpenAPI / SDK generation: agent-capability default values (#2949), a duplicate oauth2 requirement that blocked SDK generation (#2956), review-comment descriptions (#2967), and usesOrgDefault / userOrgDefault in the create-agent response (#2964, #3026) — all @uncleSlayer.
• Integration tests: enterprise-search ITs fixed to cut LLM cost (#2924), x-request-id propagated (#2970), Notion added to the workflow (#3042 — @shrikantNemiwal1), timeout raised from 180 to 300 minutes (#3075 — @siddhantota).
• Unit tests: RSS tests no longer make real network calls (#2982); test updates (#3050 — @jatingaur18).
• Cleanup: the logs directory (#3019 — @rish664) and an unused test script (#3029 — @kaushalwaghela-art) were removed.

​

New contributors

• @kaushalwaghela-art made their first contribution in #2980
• @abdul-5844 made their first contribution in #2987
• @mdakram2002 made their first contribution in #2998
• @aesp2008 made their first contribution in #3011
• @Saneajayy made their first contribution in #3048
• @igennova made their first contribution in #3060
• @siddhantota made their first contribution in #3075

​

Highlights

​

New agent loop, now the only mode

​

Query service performance overhaul

​

Parsing and data extraction as standalone services

​

Smarter, cheaper PDF and OCR pipeline

​

OpenSearch joins the supported backends

​

Redis Streams producer survives disconnects again

​

MCP grows: Omnigent integration and new tools

​

Bulk user invites, with a review step

​

Breaking changes & upgrade notes

• Knowledge-base record groups are migrated to connectors; the migration runs as part of this release (#2650 — @tushar1245).
• Parsing and extraction now run as separate services controlled by a flag — review your deployment config before enabling (#2541, #2734).
• The env template was refreshed and the redundant GRAPH_DB_TYPE key removed from Dockerfiles; the Helm chart now defaults the message broker to Redis, deploying Kafka/ZooKeeper pods only conditionally (#2887 — @tushar1245).
• The Atlassian site URL field was removed from connector configuration (#2693).
• The PipesHub Docker image RAM limit was increased — check host sizing (#2853); the database sandbox was disabled (#2854).

​

Connectors

• Slack — personal (user-scoped) connector added (#2801 — @jatingaur18).
• Box — shared files sync (#2719 — @R0gue-one).
• Google Drive — shared-drive filter (#2620) and folder filter (#2811), plus a NoneType guard before logging (#2691).
• Gmail — indexing fix (#2748).
• Outlook — removed external-user sync (#2742) and fixed idle-closed transport on long-lived connectors (#2772 — @shrikantNemiwal1).
• Jira Data Center — handle users/permission-schema endpoints that differ across DC versions (#2704 — @darshangodase).
• GitLab — connector fixes (#2920 — @darshangodase).
• Web — crawl4ai support for higher-fidelity crawling (#2641 — @R0gue-one) with follow-up bug fixes (#2750).
• RSS — indexing fix (#2729).

​

Agents & AI

• Jira Data Center and Confluence Data Center toolsets so agents can act against self-hosted Atlassian (#2661, #2667 — @darshangodase), plus Confluence tool fixes (#2812).
• OpenAI Responses API compatibility: an initial fix (#2876 — @abhishekgml1991) plus correct token-limit truncation detection and normalization of structured responses to plain text for image descriptions, streaming, and reflection history (#2917 — @kaushal1999).
• Chat attachments: markdown and txt (#2657), DOCX and CSV (#2894 — @abhishekgml1991); images capped at 25 per message (#2655).
• Retrieved records are enriched with related records from graph edges (parent, child, attachment), carrying parent ticket/project metadata into LLM context (#2596 — @darshangodase), with a follow-up fix for to_llm_context and graph enrichment (#2840).
• Blocks now carry author and mentioned-user IDs for people-aware retrieval (#2827 — @jatingaur18) and a head tag (#2936).
• Handled the deprecated temperature field for Claude models (#2723).
• Fixed the user-ask tool schema (#2873 — @vishwjeet-pawar03), sandbox code execution (#2892), artifact creation (#2916), and the completion-gate artifact trigger (#2888).

​

Fixes

• Indexing: failures propagate to queued MD5 duplicates (#2694 — @shekharkadyan), YAML support with JSON/YAML/YML handling fixes (#2731, #2752), Unicode preserved in indexing prompts (#2686 — @TatshSiow), semantic metadata added to blob storage in the legacy parsing flow (#2768), default indexing status and reindex pagination (#2735), placeholder-record handling (#2767), reindex API for non-creators (#2733), and a fix for issue #2883 (#2890 — @kaushal1999).
• File preview: reliable PPT/PPTX-to-PDF conversion with LibreOffice and CJK fonts installed in Docker (#2711, #2713 — @TatshSiow, #2732 — @hamza426-ahm).
• Chat & UI: answer-response rendering (#2868), markdown rendering with $ signs (#2922 — @tushar1245), citation markdown (#2678), citation popover no longer closes on preview (#2923 — @Pranjal-wq), Gemini PDF upload crash when the LLM returns content as a list (#2635 — @shekharkadyan), user-query UI (#2834), chat window alignment (#2862, #2875), removed the graph/plot force-download hook (#2866), external-link icons open the source web URL (#2925), folder-upload hierarchy shown with a compact add button (#2928), embedding-model download progress in the UI (#2681), fail reason no longer shown on completed records (#2671), empty manual-indexing tooltip (#2687), connector label typo (#2700).
• Knowledge hub: app-scoped search and folder filter (#2709), shared-with-me record groups (#2688), collection record stats (#2777), agent builder shows all configured collections (#2673).
• Graph queries: replaced invalid coalesce with not_null in AQL (#2914 — @tushar1245).
• Desktop: the Electron app can now switch servers (#2865 — @abhishekgml1991).
• Cleanup: removed admin/everyone broadcast notifications (#2728), stale x-oauth-header (#2664), unused redirect-uri field (#2782), user event (#2828); invite-mail error logging fixed (#2658).

​

Performance & reliability

• Stabilized MongoDB failures in deployment (#2677 — @shekharkadyan).
• Message-broker retry and configuration support (#2626 — @shrikantNemiwal1).
• Increased file-descriptor limits to prevent EMFILE errors under load (#2843 — @anupPradhan0).

​

Security & authentication

• New token scopes and JWT generators for email verification and OTP (#2784 — @rish664, #2781 — @vanshg685).
• Authentication is skipped for expired refresh tokens instead of failing noisily (#2736).
• SMTP values are masked, with related refactoring (#2793).
• GitHub Actions hardened per GitHub’s new security guidance (#2747).

​

Developer experience & infrastructure

• Anonymous fleet metrics: new metrics collection with an increased push interval and reduced metric logging (#2644 — @rish664, #2663, #2717).
• OrgContext with AsyncLocalStorage support for request-scoped tenant context in Node.js (#2787 — @rish664).
• Refactoring: KB CRUD operations moved onto the data-source processor (#2718); constants, enums, and optional keys consolidated across frontend and backend (#2785, #2788, #2789, #2794, #2796, #2847).
• OpenAPI/spec: deprecation markers (#2665), incorrect Python-style language removed (#2666), storage doc ID in the record schema and spec (#2737, #2757).
• A release-changelog skill was added for generating these curated changelogs (#2915 — @rish664).
• CI/integration tests: per-backend approval and label selection (#2823 — @uncleSlayer), enterprise-search ITs parallelized with pytest-xdist plus determinism and race fixes (#2871, #2874, #2895, #2903, #2905, #2907), parallel Arango/Neo4j runs (#2753), Redis broker for Arango ITs (#2839), Mailpit as the SMTP sink (#2855, #2856), fail-fast helper unit tests before Docker (#2913 — @shekharkadyan), hardened storage-connector sync tests (#2910), and assorted IT env vars, creds, retries, timeouts, and fixes (#2605, #2662, #2708, #2720, #2721, #2744, #2756, #2774, #2779, #2831, #2835, #2846, #2861).
• Logging: Talon warning log fixed (#2852).
• Docs & README: uv-based install instructions (#2759), Slack approval and graph-DB docs (#2771), README updates (#2679, #2689, #2698, #2838, #2841, #2864).
• Dependency bumps via Dependabot (#2649, #2659, #2696, #2697, #2727, #2741, #2754, #2764, #2765, #2766, #2776, #2807, #2813, #2814, #2815, #2816, #2817, #2818, #2820, #2893, #2897, #2899, #2900, #2908, #2909, #2921, #2929, #2930, #2931, #2935).

​

New contributors

• @shekharkadyan made their first contribution in #2677
• @hamza426-ahm made their first contribution in #2732
• @Pranjal-wq made their first contribution in #2923

​

Highlights

​

New agent loop

​

Parsing and data extraction become standalone services

​

Smarter, cheaper PDF and OCR pipeline

​

OpenSearch integration

​

Web connector rebuilt on crawl4ai

​

Bulk user invites from CSV/Excel

​

Richer retrieval context from the knowledge graph

​

Breaking changes & upgrade notes

• Knowledge-base record groups are migrated to connectors; the migration runs as part of this release (#2650 — @tushar1245).
• Parsing and extraction now run as separate services controlled by a flag — review your deployment config before enabling (#2541, #2734).
• The Atlassian site URL field was removed from connector configuration (#2693).
• The PipesHub Docker image RAM limit was increased — check host sizing (#2853).

​

Connectors

• Slack — personal (user-scoped) connector added (#2801 — @jatingaur18).
• Box — shared files sync (#2719 — @R0gue-one).
• Google Drive — shared-drive filter (#2620) and folder filter (#2811), plus a NoneType guard before logging (#2691).
• Gmail — indexing fix (#2748).
• Outlook — removed external-user sync (#2742) and fixed idle-closed transport on long-lived connectors (#2772 — @shrikantNemiwal1).
• Jira Data Center — handle users/permission-schema endpoints that differ across DC versions (#2704 — @darshangodase).
• RSS — indexing fix (#2729).

​

Agents & AI

• Jira Data Center and Confluence Data Center toolsets so agents can act against self-hosted Atlassian (#2661, #2667 — @darshangodase), plus Confluence tool fixes (#2812).
• MCP: scope fix and a new fetch-full-record API (#2726 — @uncleSlayer), server updated to v2.2.0 (#2780), and a fetch-full-record tool bug fix (#2743).
• Chat attachments: markdown and txt files supported (#2657); images capped at 25 per message (#2655).
• Handled the deprecated temperature field for Claude models (#2723).
• Blocks now carry author and mentioned-user IDs for people-aware retrieval (#2827 — @jatingaur18).

​

Fixes

• Indexing: failures now propagate to queued MD5 duplicates (#2694 — @shekharkadyan), YAML support with indexing bug fixes and proper JSON/YAML/YML handling (#2731, #2752), Unicode text preserved in indexing prompts (#2686 — @TatshSiow), semantic metadata added to blob storage in the legacy parsing flow (#2768), default indexing status and reindex pagination (#2735), connector placeholder-record handling (#2767), reindex API now works for non-creators (#2733).
• File preview: reliable PPT/PPTX-to-PDF conversion with LibreOffice and CJK fonts installed in Docker (#2711, #2713 — @TatshSiow, #2732 — @hamza426-ahm).
• Chat & UI: citation markdown rendering (#2678), Gemini PDF upload crash when the LLM returns content as a list (#2635 — @shekharkadyan), user-query UI (#2834), chat window alignment (#2862), embedding-model download progress shown in the UI (#2681), fail reason no longer shown on completed records (#2671), empty manual-indexing tooltip (#2687), connector label typo (#2700).
• Knowledge hub: app-scoped search and folder filter (#2709), shared-with-me record groups (#2688), collection record stats added (#2777), agent builder now shows all configured collections (#2673).
• Cleanup: removed admin/everyone broadcast notifications (#2728), stale x-oauth-header (#2664), unused redirect-uri field (#2782), user event (#2828); invite-mail error logging fixed (#2658).

​

Performance & reliability

• Stabilized MongoDB failures in deployment (#2677 — @shekharkadyan).
• Message-broker retry and configuration support (#2626 — @shrikantNemiwal1).
• Increased file-descriptor limits to prevent EMFILE errors under load (#2843 — @anupPradhan0).
• Database sandbox disabled (#2854).

​

Security & authentication

• New token scopes and JWT generators for email verification and OTP (#2784, #2781 — @rish664, @vanshg685).
• Authentication is skipped for expired refresh tokens instead of failing noisily (#2736).
• SMTP values are masked, with related refactoring (#2793).

​

Developer experience & infrastructure

• Anonymous fleet metrics: new metrics collection with an increased push interval and reduced metric logging (#2644 — @rish664, #2663, #2717).
• OrgContext with AsyncLocalStorage support for request-scoped tenant context in Node.js (#2787 — @rish664).
• Refactoring: KB CRUD operations moved onto the data-source processor (#2718); constants, enums, and optional keys consolidated (#2785, #2788, #2789, #2794, #2796).
• OpenAPI/spec: deprecation markers added (#2665), incorrect Python-style language removed (#2666), storage doc ID added to the record schema and spec (#2737, #2757).
• CI/integration tests: per-backend approval and label selection (#2823 — @uncleSlayer), parallel Arango/Neo4j runs (#2753), Redis broker for Arango ITs (#2839), Mailpit as the SMTP sink (#2855, #2856), GitHub Actions security hardening (#2747), and assorted IT env vars, creds, retries, timeouts, and fixes (#2662, #2708, #2720, #2721, #2744, #2756, #2774, #2779, #2831, #2835, #2846, #2861, #2605).
• Logging: Talon warning log fixed (#2852).
• Docs & README: uv-based install instructions (#2759), Slack approval and graph-DB docs (#2771), README updates (#2679, #2689, #2698, #2838, #2841, #2864).
• Dependency bumps via Dependabot (#2649, #2659, #2696, #2697, #2727, #2741, #2754, #2764, #2765, #2766, #2776, #2807, #2813, #2814, #2815, #2816, #2817, #2818, #2820).

​

New contributors

• @shekharkadyan made their first contribution in #2677
• @hamza426-ahm made their first contribution in #2732

​

Highlights

​

Jira and Confluence Data Center support

​

Agents gain action tools

​

Open model layer: OpenRouter, LiteLLM proxy, LM Studio

​

Document parsing overhaul

​

Confluence content pipeline rebuilt on blocks

​

GitLab connector matures

​

In-app notification service

​

Connectors

• Slack — new/reworked Slack connector landed (#2272 — @Ayushman444).
• Jira (Cloud) — application-role fallback (#2382), email-visibility handling (#2390), graceful handling of permission 403s (#2413), and metadata enrichment including full-record fetch (#2530, #2557 — @darshangodase).
• Confluence (Cloud) — email visibility and pagination (#2399), access option changed from checkbox to select (#2430), space-permission handling (#2420), and attachment streaming fix (#2568).
• GitLab — EE fixes (#2369), filter search using filter groups for permissions (#2371), filter-selection sync fix (#2398), auditor changes (#2417), and merge-request block-index fix with increased timeout (#2629).
• Linear — API-token authentication support (#2622 — @darshangodase).
• Notion — sync issue fixes (#2351).
• Outlook — attachment mimetype fix (#2419).
• Local filesystem / desktop app — updated sync strategies with Electron enhancements (#2403 — @Kushagra1122), assorted fixes (#2407), and an admin-access prompt in the native app during connector setup (#2447).
• Web — full-sync fix (#2584).
• S3 — health check added (#2456).

​

Agents & AI

• Citation resolution fix plus prompt improvements (#2400 — @kaushal1999).
• Fixed a token-limit blowout where whole tables were sent to the LLM (#2418).
• Fetch-URL tool made explicit for agents (#2395); sort-key bug fixed in the agent retrieval tool (#2397).
• Jira tools now check custom fields (#2591); fixed empty tool args and Confluence CQL handling (#2619).
• listAgents API now exposes toolsets and knowledge bases (#2510); MCP dependency updated to v1.2.0 (#2526).
• Removed the index-videos filter so video records are no longer excluded (#2612).

​

Fixes

• Frontend/UX: collections deletion error handling (#2353), nested record-group reindex (#2360), manual/failed reindex button in All Records (#2381), reindex options (#2423), full-sync/force-reindex confirmations (#2474), connector-instance toast and redirect naming (#2372), connector-config back buttons and confirm dialogs (#2392), connector filter modal with default filter (#2465), connector page stats handling and other page issues (#2442, #2583), sidebar apps with no children (#2436), toolset tab navigation and banner icon (#2523, #2575), collection-share owner role and role dropdown (#2425), org-name splitter (#2491).
• Chat & files: improved markdown rendering in chat (#2354), file drop zone plus PDF-citation blinker (#2505), file-upload fixes (#2483), storage mimetype list (#2567), zoom-level lock in document preview (#2517 — @rachit367), CJK input no longer replaces English words in chat (#2640), agent toolset message (#2564) and agent model selection (#2572).
• Backend: conversation pagination (#2375), KB upload API (#2396), crypto.randomUUID fallback for upload item IDs (#2391 — @TatshSiow), removed over-strict email checks (#2462), friendlier validation errors (#2535), enterprise-search error handling (#2524), fallback screen when the Node.js service is down (#2388), data-message fix (#2404) and temporary streaming-API patch (#2405), frontend API base URL pointing at the backend (#2507).
• Data layer: get_records_by_record_group Arango query (#2468), batch_update_nodes bind vars and transaction query call (#2653), Neo4j update-connector method mismatch (#2134), indexing-service record update operation (#2563), reconciliation/replace fix (#2581).
• Parsing edge cases: image handling (#2617), block data passed as none (#2633), empty image alt text in HTML-derived markdown (#2648), block splitter with image tags stripped from table headers (#2642).
• API cleanup: get-all-connectors API cleaned up (#2385) and replaced with the knowledge-hub API on the frontend (#2573); user-graph API removed from the frontend with team API fixes (#2481).

​

Performance & reliability

• Memory-leak fixes across services (#2408 — @abhishekgml1991) and a latency fix (#2602).
• New ArangoDB indexes for query performance (#2441 — @tushar1245), plus faster connector-stats and knowledge-hub search/filter queries (#2453, #2455).
• Embedding deletion now uses Qdrant’s delete-points API instead of scroll, and stale embeddings are cleaned up after record deletion (#2512 — @kaushal1999).
• Fixed a Redis deadlock on unparseable messages (#2362) and an etcd watcher leak in close() (#2367); race condition fixed in onboarding steps (#2361).
• Increased Neo4j and Qdrant timeouts (#2578); health checks moved into the Python services (#2529).
• All Records sidebar connectors now lazy-load (#2448); SQL access migrated to asyncpg (#2318 — @uncleSlayer).

​

Security & authentication

• Users created via JIT provisioning are now marked logged-in (#2349).
• Access tokens are refreshed after an OAuth revoke (#2473), and OAuth errors now propagate to the frontend instead of failing silently (#2574).
• Batch of dependency security fixes (#2593).

​

Developer experience & infrastructure

• One-command installation script for self-hosting (#2590 — @abhishekgml1991) and Dockerfile updates (#2598).
• OpenAPI/SDK: x-pipeshub-sdk extension introduced, tuned, and later removed from the public spec (#2334, #2476, #2630, #2643); spec fixes for OAuth app routes, web-search providers, backward compatibility, and breaking-change cleanup (#2249, #2479, #2646, #2652, #2637); KB upload APIs merged into one (#2570); model-type list endpoint removed (#2477); unnecessary agent routes removed (#2493).
• Zod validation layers added across attachments, conversations, agents, agent conversations, teams, and chat-mode enums (#2384, #2386, #2421, #2440, #2429, #2464, #2467, #2600 — @uncleSlayer, @realaashil, @darshangodase).
• Test coverage: Python unit tests (#2377, #2376, #2374, #2604, #2607, #2614); integration tests for user management, Jira, Confluence, KB CRUD, toolsets, text files, and agent conversations (#1863, #2151, #2565, #2504, #2632, #2601, #2647, #2651); Playwright e2e tests added to CI (#2411).
• CI plumbing: Slack reporting of IT results (#2364, #2368, #2460, #2463), workflow and environment fixes (#2373, #2412, #2415, #2437, #2452, #2480, #2582, #2585, #2592, #2595, #2599, #2609, #2610, #2615, #2623, #2636, #2548, #2566).
• Observability: request and thread IDs on services (#2611), graph-provider log cleanup (#2424), health-check 200 logs hidden (#2534), better Redis logging of unknown messages (#2536).
• Docs: README updates and badges with i18n (#2539, #2639, #2492, #2544, #2556), MariaDB deps documented in CONTRIBUTING (#2509 — @Ayushman444).
• Dependencies: MariaDB client driver v2 (#2555), uuid pinned for CommonJS compatibility (#2506), etcd3 installed from a maintained fork for protobuf 4 (#2606), and routine Dependabot bumps (#2488, #2487, #2490, #2489, #2486, #2553, #2552, #2551, #2561, #2571, #2562, #2543, #2579, #2586, #2587, #2588, #2558, #2560, #2434, #2393, #2533, #2525, #2625).

​

New contributors

• @TatshSiow made their first contribution in #2391
• @rachit367 made their first contribution in #2507

​

Highlights

​

GitLab self-managed (EE) support

​

Slimmer, simpler deployment

​

More flexible authentication

​

Connectors

• OneDrive toolset fix (#2224 — @jatingaur18).
• Arango cleanup on connector-instance deletion fixed (#2344 — @shrikantNemiwal1).
• Scheduled crawling functionality (and its tests) removed from connectors (#2333 — @Kushagra1122).

​

Agents & AI

• Deep agent no longer drags previous conversations into a new one (#2332 — @vishwjeet-pawar03).
• Model selection for agents fixed (#2336) and agents now show a creator tag (#2331 — @vishwjeet-pawar03).

​

Fixes

• File preview moved to a split-screen layout (#2339 — @vishwjeet-pawar03), recent chats are collapsible (#2328), and a workspace overflow issue was fixed (#2343).
• User-group rename API is only called when the name actually changed (#2330 — @realaashil).
• A missing method was restored (#2347 — @abhishek-pipeshub).

​

Performance & reliability

• Redis consumers drop unparsable messages completely instead of retrying them forever (#2335 — @realaashil).
• GitHub star count is cached instead of fetched per request (#2348 — @abhishekgml1991).

​

Developer experience & infrastructure

• Electron CI follow-up (#2329 — @realaashil).

​

Highlights

​

Atlassian Data Center support

​

Smaller Docker images

​

Connectors

• Local filesystem records expose a localFsRelativePath in file-record and service response types (#2311 — @Kushagra1122).
• Filters removed after a connector instance is created are now pruned (#2324 — @abhishekgml1991); non-default datetime filters no longer get selected on their own (#2317 — @shrikantNemiwal1).

​

Agents & AI

• Removing a tool from an agent is now handled cleanly (#2309 — @jatingaur18).
• Agent-builder node height no longer balloons with large system prompts (#2310 — @vishwjeet-pawar03).

​

Fixes

• Document handling: PPTX format detection (#2305 — @kaushal1999) and Mermaid diagram rendering (#2316 — @vishwjeet-pawar03).
• Chat UI: image display overflow (#2307) and a “show more” control on citations (#2308 — @vishwjeet-pawar03).
• Forgot-password now displays errors (#2315 — @vanshg685).

​

Developer experience & infrastructure

• Electron CI added for the desktop app (#2312 — @realaashil).
• Dependency bumps (#2296, #2263, #2319), with the tar bump reverted (#2322 — @abhishek-pipeshub).

​

Highlights

​

Production-ready Kubernetes deployment

​

Chat attachments and richer model support

​

A CLI and an SDK-ready API surface

​

Connectors

• SharePoint — toolset added with a shared reference-data interface (#1645 — @darshangodase), filters pagination fixed (#2182 — @R0gue-one).
• GitLab — code-file indexing fix (#2257 — @abhishek-pipeshub), EE user-email visibility fix (#2269), and connector filters (#2291 — @abhishekgml1991).
• Web connector — personal web connector fix (#2214 — @R0gue-one).
• OAuth apps — named OAuth app support (#2149 — @R0gue-one) and a toast when no OAuth app is available (#2212).
• Sync & configuration — filter changes trigger full sync (#2200 — @shrikantNemiwal1), sync schedule fixed (#2260), default filters fixed (#2251), manage-configuration button relocated (#2205), helper text when no connector is selected (#2250 — @vishwjeet-pawar03).

​

Agents & AI

• Agents know who’s asking: caller name/email passed to service agents (#2215 — @vishwjeet-pawar03) with user-context fixes for service accounts (#2258), and the current time is sent to the LLM in the user’s timezone (#2199).
• Agent UX: new creation dialog (#2244 — @jatingaur18), streaming fixed (#2248), auth name field in toolset setup (#2204).
• Guardrails: AI models and connectors can’t be deleted while an agent depends on them (#2166 — @darshangodase).
• Tools: Confluence tools now retrieve in parallel (#2233 — @vishwjeet-pawar03); the fetch-weburl tool avoids login-walled pages (#2242 — @abhishekgml1991); Redshift toolset schema tool, forced SQL routing, and asyncio fixes (#2294, #2295, #2300 — @Ayushman444); query-block call fix (#2013 — @rish664).
• Citations guarded against missing data before .get access (#2284 — @Ayushman444); internal ids stripped from Neo4j responses (#2280 — @realaashil).
• Speech recognition extended to Brave and Edge browsers (#2297 — @Kushagra1122).

​

Fixes

• Citations: scrolling and fuzzy matching for Markdown files (#2203 — @vishwjeet-pawar03), citation bubbles are links (#2290).
• All-records: scoped search/filter (#2247 — @shrikantNemiwal1), filter param persistence (#2239), search-clear behavior (#2255); collections page defaults to the first collection (#2202 — @R0gue-one); record preview rerouted to record/:id (#2235 — @vanshg685).
• Chat: input removed from shared chats (#2266 — @jatingaur18), web-search filter clearing (#2292), CSV row rendering (#2298), archive-chat Mongo query made DocumentDB-compatible (#2273 — @jatingaur18).
• Errors surfaced properly: API error messages propagate to UI toasts (#2209 — @shrikantNemiwal1), invite-mail and cancel-invite error codes corrected (#2219 — @vanshg685), reset-password “Invalid URL” fixed (#2176).
• User groups API and frontend call fixed (#2234 — @realaashil); service-page links hidden from non-admins (#2246 — @rish664); onboarding confirmation dialog (#2256); an is_external field added to org-fetching methods (#2090 — @jatingaur18); blob-storage URL fix (#2301 — @abhishekgml1991).

​

Performance & reliability

• Knowledge-hub memory-exceed fix (#2259 — @shrikantNemiwal1).
• Rate limit increased (#2236 — @abhishekgml1991).
• Frontend proactively refreshes tokens before expiry (#2240 — @tushar1245).

​

Security & authentication

• Secrets are masked in configuration API responses (#2293 — @vishwjeet-pawar03).
• MCP scopes updated for the new MCP server (#2226 — @realaashil).
• Missing SAML certificate now logs a proper error (#2206 — @vanshg685).

​

Developer experience & infrastructure

• Parsing dependencies slimmed: html2text replaced with markdownify (#2276) and ocrmypdf support removed (#2277 — @abhishekgml1991).
• Event definitions separated (#1917 — @realaashil) and the legacy baseArangoService usage removed (#2187 — @tushar1245).
• CI/integration tests: SharePoint workflow with site-name variable (#2198, #2207 — @R0gue-one), Jira credentials (#2230 — @darshangodase), auth-route ITs (#1869), storage suites ordered last (#2225), daily midnight-IST runs (#2288 — @realaashil), GCS_BUCKET_NAME variable (#2289 — @Kushagra1122), Python coverage threshold to 91.5 (#2237), test fix (#2285).
• Docs: new README with linked images (#2208, #2211 — @rish664) and an FAQ section (#2243 — @meichuanyi).
• Dependency bumps (#2195, #2227, #2228, #2252).

​

New contributors

• @meichuanyi made their first contribution in #2243

​

Highlights

​

The frontend is now Next.js

​

Coding sandbox, artifacts, and image generation

​

New connectors: GitLab, Salesforce, Zoom, personal Outlook

​

Toolsets become first-class

​

Web search and deeper agent modes

​

Indexing reconciliation and SQL connectors

​

More ways to run PipesHub

​

Breaking changes & upgrade notes

• The legacy React frontend is removed; the Next.js frontend is the only UI (#2169).
• Python services moved to Python 3.12 (#1755).
• The frontend consolidates on a single backend URL; redundant URL settings were removed (#1780).
• The SMTP step was removed from onboarding (#2180).
• Unnecessary container ports are no longer exposed, alongside a logs cleanup (#1890).
• The backend switched to the new custom MCP server (#2185).
• The dev Docker compose now defaults to Neo4j as the graph store (#2170).

​

Connectors

• SharePoint — full-sync fix, new-UI fixes, and site filters now apply to site groups (#1736, #2056, #2103 — @R0gue-one).
• Google Drive & Gmail — fixes for the new UI (#2088) and the Drive toolset (#2135).
• Jira & Confluence — multiple-site selection fixed, with a fallback (#1964, #2014); Jira and Linear files full-sync fixes (#2015); record-group/folder/record reindexing plus Linear and Zammad documentation links (#2020 — @darshangodase); email-visibility doc links on config pages (#2189).
• GitHub — assorted fixes (#1728 — @jain-ayush-wyld).
• Nextcloud — Python 3.10 typing compatibility (#1722).
• Microsoft 365 — group filtering (#1843).
• Web connector — visited-URL tracking fix (#1671), pages filter removed (#1750), sync-settings and config-page fixes (#2087, #2107).
• Object storage (S3, GCS, Azure Blob/Files) — record timestamps (#2083 — @Kushagra1122), team-app and personal S3 permission fixes (#1559 — @shrikantNemiwal1), S3-compatible blob storage support (#2154).
• Connector platform — lifecycle management supports pre-existing resources (#1971); instance removal with optimistic deletion handling (#1999, #2037, #2040 — @Kushagra1122); custom per-connector config-page logic (#2011 — @R0gue-one); sync fixes (#1980, #2001, #2005); sync strategy selector and filter dropdowns (#1998); sync filter chips (#2017); connector rename (#2110); info message and sync-confirm modal (#2117); connectors page UI (#2132); connector id display (#2167); connector icons (#1995); docs link in the config header (#2146); filter API fix (#2173).
• Connector OAuth — non-admin users can authenticate their own connectors (#2097, #2034); OAuth flows use config authorizeUrl/tokenUrl instead of the registry (#1591 — @shrikantNemiwal1); OAuth scopes reduced (#2029 — @jain-ayush-wyld); required checkboxes must be checked (#2098).
• Internal records are excluded from indexing (#1724 — @R0gue-one).

​

Agents & AI

• LLM providers: MiniMax added as a first-class provider (#1784 — @octo-patch); Ollama model updates no longer hang when no API key is set (#1923 — @JiwaniZakir); role=None validation errors in streaming chunks handled (#1785 — @Sangampal11); Azure AI setup hints with session-invalidate recovery (#2028); langchain Google AI package upgraded (#1867).
• AI-model management: model add/edit functionality (#1968 — @vanshg685), model name display and rendering visibility (#1961, #1896), settings shortcut in the models box (#2044), optional embedding-model step at onboarding (#1992), fallback to the default model when a conversation’s model is unavailable (#2116 — @tushar1245), agent chat restricted to agent-configured models (#1942), model selection fixes for chats (#2089).
• Agent platform: sharing (#1604), sidebar toolset registry (#1725), canvas fixes (#1898), agent modes in chat (#1912), knowledge retrieval (#1916, #2081), non-OAuth toolsets (#1845), knowledge/tool filters (#2016, #2144, #2145), react-agent handoff on direct answers (#2036), parallel conversation handling (#2175), duplicate streaming and citation hover states (#2060), chat archive and rename (#2051), capability-summary prompt with SQL as an internal domain tool (#2093), one instance of a toolset per agent (#2101), system prompt fix (#1941), agent migration removed (#2133).
• Agent service accounts added (#1767 — @vishwjeet-pawar03) with dialog and behavior fixes (#1865, #1934).
• Slack: toolset DM queries (#2104), response id and timestamp resolution (#2129 — @Ayushman444), bot follow-ups in agents (#2127), link previews (#1944), time context across Slack, assistant, and agents (#1881 — @kaushal1999).
• Citations: refactor with cleanup and bug fixes (#1840 — @kaushal1999), grouping plus CSV support (#1976), consistent single-citation format (#2003, #2021), malformed and bare citation handling (#2159), citation dropping in multi-hop retrieval fixed (#2067), node type in citation URLs (#1991), text-fragment URL rewrite guard (#1821), duplicate popovers prevented (#2121), full-screen highlight fix (#1957), Python-side citation bug fix (#1831).
• Streaming: word-by-word streaming instead of large chunks (#2141 — @kaushal1999), chat stream fix (#2136).
• Knowledge search & graph: KB filter in knowledge search (#1779), result fixes (#1895, #2092), knowledge-hub parent permissions (#1814), document scopes removed and stream_record fixed (#1667), parent-child edge update (#1532 — @jatingaur18).
• SQL toolsets: query fixes (#2082), downloadable query results (#2006).

​

Fixes

• Chat UX: search and unsaved-changes dialog (#1943), share/archive fixes (#1951, #2078), time grouping (#2046), scrolling during streaming (#2181, #2186), conversation margins (#1962), filter persistence (#2099), collection filter fixes (#2061, #2071, #2142), agent chats separated from the assistant page (#2130, #2140), three-dots menu reload (#2065), archive restore/delete placement (#2106), chat feedback (#2113, #2118), hardcoded suggestions and follow-ups hidden (#1967), team creation from chat share (#1979).
• All-records & collections: child-record traversal (#2007), filters handling (#2047, #2049, #2111), sidebar and nested-node pagination (#2077, #2084), search results clickable (#2156), connector links (#1993), record detail page (#2019, #2025), record status reason visibility (#2079), collection page fixes (#1956, #2086).
• Files & preview: upload API paths and versioning for local + S3 storage (#1505 — @Ayushman444), folder upload (#2161), upload status (#2009), Markdown uploads in the new UI (#1977), DOCX rendering with resizable preview (#2045), PDF viewer fixes (#2094, #2124), Excel viewer (#2163), web-URL placement in preview details (#2155), open-in-web button (#2096), large queries / file preview / web-URL fixes (#2068), upload-docs scroll bar with citation format (#2021).
• Workspace, teams & users: workspace UI fixes (#1935, #2057, #1982, #2064 — @akshay-pythm), action menu hidden while a sub-menu is open (#1907), pagination reset prevented (#1906), team pages and route guard (#2066, #1989), draggable sidebar with persistent width (#1893 — @akshay-pythm), infinite scrolling in users/teams/shared chats (#1953), user list filters with generic messaging (#2114), unique key fix (#1892).
• Profile, org & onboarding: profile and logo fixes (#1904, #2035, #1938), invite fixes with resend/cancel options (#1894, #1986), language selection at sign-up (#2043), default-group deletion guard (#2032).
• Internationalization: i18n across workspace pages (#1996, #2004 — @akshay-pythm), missing translations added (#1952), key fixes (#2063), unused translations removed (#2150).
• UI polish: loaders and spinners on actions (#1940, #1966, #2002), infra service logos and icon fixes (#1972, #2023, #2109), auth buttons (#2152), toolset UI bugs (#1963), agent chat view button (#1974), wispr drawer (#2039), attachment button hidden until ready plus favicon (#1975), href support in UI components (#2052), form validations (#2070), number filters saved as numbers (#2112), assorted minor UI fixes (#2128, #2095, #2080, #2018, #2131).
• Misc: error propagation to frontend toasts (#2153), toolset redirect on Docker (#1994), frontend build failure fix (#1921).

​

Performance & reliability

• Indexing throughput and Docling worker tuning (see Highlights) plus latest-content MD5 dedup (#1770).
• Retry on ArangoDB deadlocks for data-source entities (#1864 — @tushar1245), an optimized KB deletion query (#2184 — @shrikantNemiwal1), and an app-permission subquery fix (#1856).
• Config reads no longer served from cache, avoiding stale credentials (#2126 — @kaushal1999).
• Health-check race condition fixed (#2162), interval moved to 10 minutes (#2008), health checks with user full-name validation (#1978) and a health-page revamp (#1988).
• Frontend memory optimized to prevent OOM crashes (#1766 — @anupPradhan0) and auto-restart if the frontend goes down (#1757).
• Frontend API timeout raised to 60s (#2122); aiohttp content-type check (#2183 — @Ayushman444).

​

Security & authentication

• Authentication flow changes (#1723 — @vanshg685) and email-change verification via mail (#1748, #1936).
• Reset-password always returns 200 with a generic message to prevent account enumeration (#2041 — @realaashil).
• Blocked users get a 24-hour cooling period (#2062 — @realaashil).
• Frontend URL localhost validation (#1825), vulnerable unused package removed (#2000), deployment key added to the unencrypted config list (#2076).
• SAML and default embedding-model frontend fixes (#2105).

​

Developer experience & infrastructure

• Unit tests pushed above 90% coverage on Node.js and Python services (#1740, #1762, #1790, #1806, #1853, #1969, #2123) with threshold/fix churn (#1830, #1834, #1854, #2069) and weekly coverage reporting to Slack (#1744, #1754, #1836, #1837, #1838).
• Integration-test infrastructure built out across Arango, storage, Confluence, MCP, and OpenAPI suites (#1727, #1729, #1732, #1733, #1734, #1735, #1737, #1778, #1861, #1866, #1871, #1872, #1879, #1908, #1919, #1922, #1965, #1990, #2157, #2160, #2164, #2174) with CI runner sizing and secrets (#1826, #1900, #1901).
• Frontend test tooling: Playwright integration (#1958), Monocart coverage reports (#1959), Claude Code review workflow (#1955, #1973, #1985, #1987, #1981).
• Validation-failure logging was made a warning, then reverted after review (#1849, #1860).
• Docs: README refresh with TOC (#1800, #1803, #1888, #2177), SDK/MCP repo links (#1777, #1798), contributing guide with dependabot fixes (#2171).
• Cleanup: lint fixes (#1721), noisy logs removed or demoted (#1745, #1771, #1808), opik package removed (#1753), unused Docling files deleted (#1764), Neo4j Makefile (#1817), email templates (#2010), Docker compose/YAML fixes (#1793, #1850), missing lib folder (#1887).
• Dependency bumps via Dependabot across Python, Node.js, and the frontends (#1760, #1761, #1768, #1769, #1775, #1776, #1783, #1786, #1791, #1810, #1811, #1816, #1818, #1827, #1847, #1851, #1852, #1858, #1859, #1868, #1880, #1883, #1884, #1886, #1909, #1910, #1924, #1925, #1928, #1930, #1933, #1946, #1947, #1948, #1949, #1950, #2026, #2027, #2030, #2055, #2058, #2120, #2172), including one axios revert (#1983).

​

New contributors

• @octo-patch made their first contribution in #1784
• @Sangampal11 made their first contribution in #1785
• @akshay-pythm made their first contribution in #1893
• @JiwaniZakir made their first contribution in #1923

​

Highlights

​

Next.js frontend preview

​

Coding sandbox, artifacts, and image generation

​

New connectors: GitLab, Salesforce, Zoom, personal Outlook

​

Toolsets become first-class

​

Web search and deeper agent modes

​

Indexing reconciliation, SQL connectors, and deployment flexibility

​

Breaking changes & upgrade notes

• Python services moved to Python 3.12 (#1755).
• The frontend consolidates on a single backend URL; redundant URL settings were removed (#1780).
• Unnecessary container ports are no longer exposed, alongside a logs cleanup (#1890).

​

Connectors

• SharePoint — full-sync fix, new-UI fixes, site filters apply to site groups (#1736, #2056, #2103 — @R0gue-one).
• Google Drive & Gmail — new-UI fixes (#2088) and Drive toolset fix (#2135).
• Jira & Confluence — multiple-site selection fixed with a fallback (#1964, #2014); Jira and Linear files full-sync fixes (#2015); reindexing for record groups/folders/records plus Linear and Zammad doc links (#2020 — @darshangodase).
• GitHub — assorted fixes (#1728 — @jain-ayush-wyld); Nextcloud — Python 3.10 typing compatibility (#1722); Microsoft 365 — group filtering (#1843).
• Web connector — visited-URL tracking fix (#1671), pages filter removed (#1750), sync-settings and config-page fixes (#2087, #2107).
• Object storage (S3, GCS, Azure Blob/Files) — record timestamps (#2083 — @Kushagra1122), team-app and personal S3 permission fixes (#1559 — @shrikantNemiwal1), S3-compatible blob storage support (#2154).
• Connector platform — lifecycle support for pre-existing resources (#1971); instance removal with optimistic deletion (#1999, #2037, #2040 — @Kushagra1122); custom config-page logic (#2011 — @R0gue-one); sync fixes (#1980, #2001, #2005); sync strategy selector (#1998) and filter chips (#2017); rename (#2110); info message and sync-confirm modal (#2117); page UI (#2132); icons (#1995); docs link in config header (#2146).
• Connector OAuth — non-admin users can authenticate their own connectors (#2097, #2034); OAuth uses config authorizeUrl/tokenUrl (#1591 — @shrikantNemiwal1); scopes reduced (#2029 — @jain-ayush-wyld); required checkboxes enforced (#2098).
• Internal records are excluded from indexing (#1724 — @R0gue-one).

​

Agents & AI

• LLM providers: MiniMax added (#1784 — @octo-patch); Ollama update no longer hangs without an API key (#1923 — @JiwaniZakir); role=None streaming chunks handled (#1785 — @Sangampal11); Azure AI hints with session recovery (#2028); langchain Google package upgraded (#1867).
• AI-model management: add/edit models (#1968 — @vanshg685), model name display (#1961, #1896), settings shortcut (#2044), optional embedding-model onboarding step (#1992), default-model fallback (#2116 — @tushar1245), agent chat restricted to configured models (#1942), model selection fixes (#2089).
• Agent platform: sharing (#1604), sidebar toolset registry (#1725), canvas (#1898), modes (#1912), knowledge retrieval (#1916, #2081), non-OAuth toolsets (#1845), knowledge/tool filters (#2016, #2144, #2145), react-agent handoff (#2036), duplicate streaming and citation hover states (#2060), chat archive/rename (#2051), capability prompt with SQL as internal tool (#2093), single toolset instance per agent (#2101), system prompt fix (#1941), agent migration removed (#2133).
• Agent service accounts (#1767 — @vishwjeet-pawar03) with dialog and behavior fixes (#1865, #1934).
• Slack: toolset DM queries (#2104), id/timestamp resolution (#2129 — @Ayushman444), bot follow-ups (#2127), link previews (#1944), time context (#1881 — @kaushal1999).
• Citations: refactor with cleanup (#1840 — @kaushal1999), grouping and CSV support (#1976), consistent format (#2003, #2021), multi-hop dropping fix (#2067), node type in URLs (#1991), text-fragment guard (#1821), popover and highlight fixes (#2121, #1957), Python-side bug fix (#1831).
• Streaming: word-by-word streaming (#2141 — @kaushal1999), chat stream fix (#2136).
• Knowledge search & graph: KB filter (#1779), result fixes (#1895, #2092), parent permissions (#1814), document scopes removed and stream_record fixed (#1667), parent-child edge update (#1532 — @jatingaur18).
• SQL toolsets: query fixes (#2082), downloadable query results (#2006).

​

Fixes

• Chat UX: search and unsaved-changes dialog (#1943), share/archive fixes (#1951, #2078), time grouping (#2046), margins (#1962), filter persistence (#2099), collection filters (#2061, #2071, #2142), agent chats separated from the assistant page (#2130, #2140), menu reload (#2065), archive buttons (#2106), feedback (#2113, #2118), suggestions hidden (#1967), team creation from chat share (#1979).
• All-records & collections: child-record traversal (#2007), filters handling (#2047, #2049, #2111), sidebar and pagination (#2077, #2084), clickable search results (#2156), connector links (#1993), record detail page (#2019, #2025), status reason visibility (#2079), collection page fixes (#1956, #2086).
• Files & preview: upload API paths and versioning for local + S3 (#1505 — @Ayushman444), upload status (#2009), Markdown uploads (#1977), DOCX rendering and resizable preview (#2045), PDF viewer fixes (#2094, #2124), web-URL placement (#2155), open-in-web button (#2096), large queries / preview / web-URL fixes (#2068).
• Workspace, teams & users: workspace UI fixes (#1935, #2057, #1982, #2064 — @akshay-pythm), action menu (#1907), pagination reset (#1906), team pages and route guard (#2066, #1989), draggable sidebar (#1893 — @akshay-pythm), infinite scrolling (#1953), user filters (#2114), unique key (#1892).
• Profile, org & onboarding: profile/logo fixes (#1904, #2035, #1938), invites with resend/cancel (#1894, #1986), sign-up language selection (#2043), default-group deletion guard (#2032).
• i18n: workspace pages (#1996, #2004 — @akshay-pythm), missing translations (#1952), key fixes (#2063), unused translations removed (#2150).
• UI polish: loaders/spinners (#1940, #1966, #2002), logos/icons (#1972, #2023, #2109), auth buttons (#2152), toolset UI bugs (#1963), agent chat view button (#1974), wispr drawer (#2039), attachment button hidden plus favicon (#1975), href support (#2052), form validations (#2070), number filters saved as numbers (#2112), assorted minor fixes (#2128, #2095, #2080, #2018, #2131).
• Misc: error propagation to frontend toasts (#2153), toolset redirect on Docker (#1994), frontend build failure fix (#1921).

​

Performance & reliability

• Latest-content MD5 hashing for duplicate detection (#1770 — @kaushal1999).
• Retry on ArangoDB deadlocks (#1864 — @tushar1245) and an app-permission subquery fix (#1856).
• Config reads no longer served from cache (#2126 — @kaushal1999).
• Health checks: 10-minute interval (#2008), full-name validation (#1978), health-page revamp (#1988).
• Frontend memory optimized against OOM (#1766 — @anupPradhan0), auto-restart if the frontend goes down (#1757), API timeout raised to 60s (#2122).

​

Security & authentication

• Authentication flow changes (#1723 — @vanshg685) and email-change verification via mail (#1748, #1936).
• Reset-password always returns 200 with a generic message (#2041 — @realaashil); blocked users get a 24-hour cooling period (#2062).
• Frontend URL localhost validation (#1825), vulnerable unused package removed (#2000), deployment key added to the unencrypted list (#2076), SAML and default embedding-model fixes (#2105).

​

Developer experience & infrastructure

• Unit-test coverage pushed above 90% across Node.js and Python (#1740, #1762, #1790, #1806, #1853, #1969, #2123) with threshold/fix churn (#1830, #1834, #1854, #2069) and weekly reports to Slack (#1744, #1754, #1836, #1837, #1838).
• Integration-test infrastructure across Arango, storage, Confluence, MCP, and OpenAPI suites (#1727, #1729, #1732, #1733, #1734, #1735, #1737, #1778, #1861, #1866, #1871, #1872, #1879, #1908, #1919, #1922, #1990) with runner sizing and secrets (#1826, #1900, #1901).
• Frontend test tooling: Playwright (#1958), Monocart coverage (#1959), Claude Code review workflow (#1955, #1973, #1985, #1987, #1981).
• Validation-failure logging made a warning, then reverted (#1849, #1860).
• Docs: README refresh with TOC (#1800, #1803, #1888), SDK/MCP repo links (#1777, #1798).
• Cleanup: lint fixes (#1721), log noise removed or demoted (#1745, #1771, #1808), opik package removed (#1753), unused Docling files deleted (#1764), Neo4j Makefile (#1817), email templates (#2010), Docker compose/YAML fixes (#1793, #1850), missing lib folder (#1887).
• Dependency bumps via Dependabot across Python, Node.js, and the frontends (#1760, #1761, #1768, #1769, #1775, #1776, #1783, #1786, #1791, #1810, #1811, #1816, #1818, #1827, #1847, #1851, #1852, #1858, #1859, #1868, #1880, #1883, #1884, #1886, #1909, #1910, #1924, #1925, #1928, #1930, #1933, #1946, #1947, #1948, #1949, #1950, #2026, #2027, #2030, #2058), including one axios revert (#1983).

​

New contributors

• @octo-patch made their first contribution in #1784
• @Sangampal11 made their first contribution in #1785
• @akshay-pythm made their first contribution in #1893
• @JiwaniZakir made their first contribution in #1923

​

Highlights

​

Agent toolsets and a deep agent

​

MCP integration on a maturing OAuth2 stack

​

Slack bot: streaming, multi-bot, multi-user threads

​

Neo4j as an alternative graph database

​

Web connector overhaul

​

Seven new datasources and connectors

​

Breaking changes & upgrade notes

• Deprecated endpoints were removed from the OpenAPI spec (#1582, #1597); the signed-URL download route was removed everywhere in favor of authenticated downloads (#1470).
• The connector_disabled and paused indexing statuses were removed from records (#1663); connectors configured for manual sync no longer sync on service startup (#1592).
• The KV migration is skipped when etcd is not set up (#1562).

​

Connectors

• Google Workspace — credentials-creation fix in google.py (#1473), datasource refresh (#1487), shared-with-me query and Drive workspace-credentials fix (#1484), and folder sync for Drive and KB (#1623, #1629 — @R0gue-one)
• Outlook — group-mail web URL and indexing fixes (#1478), recipients and subject included in indexed mail body (#1493), and guidance plus occurrence-delete updates (#1647 — @jatingaur18)
• Microsoft / SharePoint — user-permission mapping fix (#1558 — @mstcomit) and Graph API pagination so all SharePoint sites are discovered (#1550 — @mstcomit)
• Box — bug and issue fixes (#1501 — @realaashil)
• GitHub — user-repos fixes for the personal connector (#1662 — @jain-ayush-wyld)
• Shared connector framework — full sync triggered on filter change with edge deletion (#1515, #1569 — @shrikantNemiwal1), sync-issue fixes (#1675), config-path fix for connector instance deletion (#1534)

​

Agents & AI

• AWS Bedrock can optionally read IAM role credentials from EC2 (#1514 — @abhishekgml1991)
• New OpenCV + PyMuPDF parser in the parsing pipeline (#1655 — @kaushal1999)
• Fall back to embedding only when document classification fails (#1575)
• Knowledge-retrieval fix (#1520) and Teams channel/team member get-and-add changes (#1646 — @vanshg685)
• Record message enhancements for indexing (#1513)

​

Fixes

• Knowledge Hub: scoped search (#1464), origin filter (#1653), KB-sharing status (#1697), KB move API (#1510), record sizeInBytes and default KB name (#1548), a query-change revert across both graph providers (#1508), and a generic KB implementation refactor (#1672)
• Graph layer: wrong argument order in update_node (#1521), collection argument removed from delete_parent_child_edge (#1621), get_user called with active as a keyword (#1702), permission-edge cleanup when record_group_id changes (#1536 — @jatingaur18), schema fix (#1588)
• Records/indexing: cross-loop indexing error (#1555 — @Ayushman444), record-stream API error handling (#1660), app-user validation (#1590)
• UI: model friendly names (#1491), manual-indexing button in connector stats (#1489), connector-stats card buttons (#1557), missing icon (#1561), connector icon for agent knowledge (#1668), axios errors surfaced via snackbar (#1690), non-editable inputs disabled (#1694)
• Misc: CORS now allows PATCH (#1693), Redis client fix (#1586)

​

Performance & reliability

• Cross-connector record leakage caused by identical virtualIds fixed (regression) (#1699 — @tushar1245)
• Knowledge Hub search/filter latency reduced (#1472), stats-query response time cut (#1574), and search_with_filters query performance improved (#1673)
• Connector instance deletion made async with query optimization (#1619); duplicate embedding deletions skipped during bulk delete on instance removal (#1517)
• Compression for local storage (#1468), cache invalidation published from Node.js (#1474), and config/notification services closed cleanly during shutdown (#1480)

​

Security & authentication

• JIT SAML SSO support (#1502 — @vanshg685) and admin ability to unblock users, with a separate SMTP settings section (#1495)
• Token refresh added to the axios request interceptor (#1656 — @tushar1245)
• Dependency upgrades: axios (#1471), express-rate-limit (#1630), undici (#1665)

​

Developer experience & infrastructure

• Integration-test workflow for PRs and scheduled runs (#1613 — @Kushagra1122), with an S3 integration test and in-memory store client property (#1523), runner supervision (#1689), and workflow fixes (#1711, #1715, #1716 — @realaashil)
• OpenAPI validator with PR sanity checks (#1482, #1486, #1500 — @rish664), spec check dropped for internal services (#1628), and OpenAPI spec fixes (#1578, #1585)
• Linting: broader rules (#1595), unused-variable and dead-code rules (#1679), lint PR checks running behind a flag (#1704), Ruff/Pyright and Node.js lint made non-fatal (#1706, #1717)
• Gemini rules added then reverted (#1678, #1714); build-break and build fixes with unused-code removal (#1580, #1664)
• Logger colors for warn/error (#1631), README update (#1637), signed-URL route cleanup (#1470)

​

New contributors

• @vanshg685 made their first contribution in #1502
• @mstcomit made their first contribution in #1558
• @Ayushman444 made their first contribution in #1555
• @anupPradhan0 made their first contribution in #1696

​

Highlights

​

15+ new connectors and datasources

​

Multi-instance connectors with filters and lifecycle management

​

Knowledge bases become connectors, plus a Knowledge Hub API

​

Graph database abstraction layer, async ArangoDB

​

PipesHub as an OAuth2.0 provider, JIT provisioning, and security hardening

​

Citations that are actually usable

​

Smarter document parsing: VLM OCR, structured output, spreadsheets

​

Breaking changes & upgrade notes

• Knowledge bases are migrated into connector form on upgrade (#1371); related KB/folder migrations run automatically (#1170, #1179, #1406, #1452).
• The deprecated “Public connector URL” setting was removed from frontend and backend (#1233, #1337).
• Individual account creation was removed (#1382); common is no longer the default Microsoft tenant ID (#1267).
• New environment variables control rate limiting and access/refresh-token lifetimes (#1255); LLM row-processing count for tables is now env-configurable (#1319); Kafka partitions default to 1 (#1384).

​

Connectors

• Jira — connector-ID fixes, dynamic project_keys filter options, web URLs on comment and attachment records, scope fixes, site URL removed from credentials, multi-project sync fix, block-group implementation, and Arango-service removal from record deletion (#1180, #1186, #1204, #1298, #1247, #1417, #1356, #1197 — @darshangodase)
• Box — file-extension filters, incremental folder sync, folder collaborations, date filters with scheduled sync as the default strategy, CCG authentication replacing manual token handling, and URL/reindexing fixes (#1190, #1240, #1277, #1283, #1275 — @SharmaNishchay)
• SharePoint & OneDrive — reindex and incremental sync, group-sync patches, record-group edges with custom site groups, and assorted bug fixes (#1177, #1305, #1362, #1385, #1442 — @R0gue-one)
• Google Workspace — individual-user sync fix, auth fix, Gmail refactor, shared-drive sync events, shared-with-me record group, drive streaming patch, and drive/frontend bug fixes (#1174, #1287, #1340, #1398, #1415, #1369, #1313)
• Confluence — user-sync premature termination fix, permission handling for users without email, and email-warning/web-URL/reindex-event fixes (#1219 — @codingiran, #1221, #1239 — @shrikantNemiwal1)
• Outlook — filter options, user groups, and default filters (#1210, #1271 — @shrikantNemiwal1)
• S3 / GCS / Azure Blob / Azure Files — rename/move and content-change detection, revision handling and file-state detection, folder-hierarchy derivation, multi-select file-extension filters, and follow-up fixes (#1297, #1342, #1363, #1429, #1426, #1332, #1355, #1254 — @Kushagra1122)
• Web — general fixes, record group support, and patches (#1229, #1379, #1392 — @R0gue-one)
• Dropbox — personal previewRenderable record-type check and folder fix (#1222, #1404)
• GitHub — starred_at fix (#1211 — @rish231294)
• Linear — fix for existing files as child records (#1443)
• Nextcloud — stop storing paths for Nextcloud records (#1437 — @realaashil)
• Workday — client moved to its correct path (#1251)
• Shared connector framework — DataSourceEntitiesProcessor derives parent external records (#1396), fix for hierarchical record structure with multi-instance connectors (#1183), scope deletion on auth change (#1351), streaming fix for attachments with special characters (#1335), and cached-credentials validation (#1160)

​

Agents & AI

• Optimized agent execution with reduced latency and improved tool handling (#1096 — @vishwjeet-pawar03), plus agent and agent-streaming fixes (#1349, #1365) and removal of non-working bind-tools code (#1149)
• Hybrid Qdrant search and token-limit handling in tool calls (#1421 — @kaushal1999)
• Custom system prompts, backend and frontend (#1161, #1230)
• Embedding model priority ordering (#1147 — @jain-ayush-wyld)
• Manual indexing, with UI display (#1173 — @R0gue-one, #1361) and record-group reindex (#1175)
• MD5-based deduplication across all record types (#1208)
• Docling BytesIO handling and mimetype-to-extension fallback in parsing (#1201, #1218)

​

Fixes

• Connector configuration: trailing spaces and whitespace at init (#1165, #1185), trailing slashes in URLs (#1164), removed 10-character minimum-length validation (#1167), filter key property and instance destroy on toggle (#1184), duplicate instance-name error message and no-subject mail sync (#1209)
• Query/retrieval: record access bug (#1182), retrieval bug (#1288), region mapper for search queries (#1425), query update (#1428), org-record filters (#1216)
• Knowledge Hub / KB: empty items list (#1301), records fetching with filters (#1306), record stream (#1333), duplicate results (#1410), KB records API (#1397), folder-creation bug (#1323), deletion query (#1402), document download (#1377, #1408), owner-role changes (#1454), records permissions with query refactoring (#1439), view-details on empty KBs/folders (#1401)
• Records & record groups: creation (#1383), processing order (#1330), groups and relations (#1391), placeholder-record connections (#1407), root record group and app edge with migration fix (#1450, #1452), record-group types (#1451), type display (#1411), isInternal field (#1431), record deletion (#1326), record-name encoding (#1224), block/block-group ordering and handling (#1405, #1416)
• Indexing: skip records from deleted connector instances (#1268, #1272), in-progress KB records failing to reindex after service restart (#1171 — @abhishekgml1991), status set to queued (#1312), clearer error message on indexing failure (#1290)
• Schema: keep deprecated Arango fields instead of removing them (#1317), entityRelation/project/link schemas added to the base service (#1353), Arango query execution/validation fix and its revert (#1445, #1447), schema-change reverts (#1393)
• UI: 404-page logo (#1196), web-URL highlighting (#1231) and hide-web-URL support (#1307), white-label logo (#1292), PDF-highlighter and chip styling (#1265), refresh after image upload (#1276), filter-settings next button (#1256), file-extension filter rendering (#1322), form autoscroll and error handling (#1381), frontend-URL/services section removal (#1263), all-records issues and pagination (#1390, #1364, #1458), record-details source header (#1399), filter cleared on record click (#1395), PPT rendering (#1459), non-admin redirect from connector routes (#1446)
• Misc: upload API payload structure (#1295), exception handling (#1296), organization permission fix (#1348), all-records count with dead-code removal (#1370), minor fixes (#1394)

​

Performance & reliability

• Kafka: producer-initialization race fixed (#1248), semaphore-based event-consumption control later decoupled from consumption (#1176, #1291), and indexing semaphore tuned to 15 (#1309)
• Record JSON size reduction — empty-field stripping and index ranges for Block/BlockGroup children — landed, was reverted, then re-landed (#1380, #1387, #1388)
• Parallel record fetching from blob storage (#1436) and minimal batch size for CPU embeddings (#1449)
• Knowledge Hub filter queries moved into ArangoDB and API queries refactored (#1412, #1457)

​

Security & authentication

• Removed emails from the getUsers API (#1234), general security bug fix (#1238), AWS region input sanitization (#1242), and input validation fix (#1253)
• Token handling: removed unnecessary token generation (#1280), auth tokens used for signed-URL record downloads (#1282), redirect-URI fix (#1246), and OAuth configuration support (#1225)
• Dependency upgrades for vulnerabilities: aiohttp (#1217), qs (#1223), config-resolver (#1243), jwk-to-pem (#1244), react-router (#1236), undici (#1308), lodash-es (#1341) with a lodash dependency fix (#1360), multer (#1423), fast-xml-parser and python-multipart (#1422), and a protobuf bump that was subsequently reverted (#1420, #1424)

​

Developer experience & infrastructure

• Helm chart for Kubernetes deployment (#1338 — @KrKOo)
• Kafka SASL support and configuration (#1359, #1357), Redis TLS support (#1367), Redis as a key-value store for config (#1281), and S3-compatible object storage as an internal storage backend (#1400)
• Consolidated, cleaned OpenAPI specs with improved docs UI (#1311, #1329), API documentation for authentication, configuration, and crawling (#1139), and docker/npm-start fixes for the spec tooling (#1325, #1327, #1328)
• CI/build: lint-job fix (#1212), env flag (#1215), missing cachetools package (#1220), environment config variable (#1462)
• Cleanup: unused domain-extraction logic (#1237), TEST_EVIDENCE.md deleted (#1250), debug-level logging in GraphDataStore (#1270), schema cleanup and ticket enum/edge-type standardization (#1320, #1259), ticketRelation schema removal (#1352), unused AWS Amplify removal (#1419), unused code removal (#1440)

​

New contributors

• @codingiran made their first contribution in #1219
• @Kushagra1122 made their first contribution in #1235
• @loganaden made their first contribution in #1252
• @roy-sid made their first contribution in #1002
• @KrKOo made their first contribution in #1338
• @realaashil made their first contribution in #1437

​

Highlights

​

Security hardening for production

​

Connector filters mature

​

SharePoint pages and Google file parsing

​

Connectors

• Google Drive — handle a missing refresh-token-expiry key during OAuth (#1123 — @abhishekgml1991).
• Trello — Trello datasource and client added (#1115 — @JustChaos10, first contribution).
• Microsoft Graph — fixed a pagination error (#1150 — @R0gue-one).
• Added a generic OAuth class for testing external connector example files (#1134 — @uncleSlayer).

​

Agents & AI

• Fixed a metadata-enhancement bug in the query pipeline (#1081 — @kaushal1999) and a bug in OpenAI-compatible embeddings (#1125).
• Multiple citations inside a single bracket pair are now handled (#1124 — @kaushal1999), and the reflection retry count was raised to 2 (#1128).

​

Fixes

• Empty-list handling for connector on_new_* methods (#1133 — @shrikantNemiwal1) and re-index no longer sends an empty MIME type to indexing (#1135).
• Fixed schema issues (#1158 — @abhishekgml1991) and added a placeholder implementation (#1159 — @R0gue-one).

​

Developer experience & infrastructure

• Telemetry and analytics: GitHub lead tracking (#1137 — @rish231294) and names added to metrics (#1138 — @rish664).

​

New contributors

• @JustChaos10 made their first contribution in #1115
• @tushar1245 made their first contribution in #1153

​

Highlights

​

Connector filters