Skip to content

Knowledge Intelligence 3.0: enterprise RAG platform - #23

Merged
Safiullah-Rahu merged 5 commits into
mainfrom
release/knowledge-intelligence-3.0
Nov 2, 2026
Merged

Safiullah-Rahu merged 5 commits into
mainfrom
release/knowledge-intelligence-3.0

Conversation

@Safiullah-Rahu

Copy link
Copy Markdown
Owner

Turns the CSV-AI Streamlit app into Knowledge Intelligence, a multi-tenant, permission-aware enterprise RAG platform (FastAPI + PostgreSQL/pgvector backend, Next.js frontend).

What's in it

  • Tenancy and access control: forced row-level security on every tenant table, RBAC with custom roles, resource grants, deny rules and expiry, an enforced org security policy (required MFA, session lifetime), and an append-only audit log.
  • Knowledge: ingestion of 30+ formats with page/sheet/row locators, hybrid dense + BM25 retrieval with reranking, and grounded answers with verified citations and refusals.
  • Data: Text-to-SQL with validation, sandboxing and approvals; database connectors with incremental sync.
  • Integrations: website, sitemap, RSS, GitHub, S3 and GitLab (live-tested), plus OAuth/token connectors for Google Drive, OneDrive, SharePoint, Dropbox, Notion, Confluence, Slack, Teams and IMAP. These are beta until each vendor's OAuth app is configured; see DEPLOYMENT.md.
  • Agents: tools inherit the user's permissions, and approved tools resume the paused run in the background.
  • Models: Anthropic (API 2026-10-21), OpenAI and others, with live model discovery and an optional platform default model.
  • UI: the "Porcelain/Graphite" design system, streaming chat with citations, and the live Connections map.

History

main held the original 2024 app, which shares no history with the newer v2 line this work is built on. The head commit is a merge whose parents are the old main and the new work, so both histories are kept and nothing is force-pushed. Its contents are exactly the new platform.

Verification

  • Backend: 200 tests pass on PostgreSQL 16 + pgvector; ruff is clean.
  • Frontend: tsc and next build pass (27 routes).
  • Browser: the Playwright E2E suite (frontend/e2e, 11 scenarios incl. mobile) passes locally.
  • Live model runs: a 36-question evaluation with Claude Sonnet 5.5, Claude Opus 5.5 and GPT-5.4 mini, and an end-to-end approval → resume run on Claude. Results are in docs/RAG_EVALUATION.md.
  • CI: the workflow now uses a pgvector service container. This PR is its first run on GitHub.

What's done, beta and planned is listed in IMPLEMENTATION_STATUS.md.

🤖 Generated with Claude Code

Safiullah-Rahu and others added 3 commits May 20, 2026 18:41
Transform the single-user CSV-AI Streamlit app into Knowledge Intelligence,
a multi-tenant, permission-aware RAG and knowledge platform. CSV-AI's data
loading, profiling and provider abstraction are kept (moved under backend/app).

- FastAPI backend on PostgreSQL 16 + pgvector: forced row-level security on
  every tenant table, RBAC (custom roles, resource grants, deny rules,
  expiry), enforced org security policy (required MFA, session lifetime),
  append-only audit log, encrypted credentials, SSRF guard
- Ingestion of 30+ formats with locators; hybrid dense/BM25 retrieval,
  reranking, grounded answers with verified citations and refusals
- Text-to-SQL with validation, sandboxing and approvals (withdrawable,
  expiring); database connectors with incremental sync
- Integrations: website, sitemap, RSS, GitHub, S3, GitLab, plus OAuth/token
  connectors for Google Drive, OneDrive, SharePoint, Dropbox, Notion,
  Confluence, Slack, Teams and IMAP (beta until vendor apps are configured)
- Agents whose tools inherit the user's permissions; approved tools resume
  the paused run in the background
- LLM providers incl. Anthropic (API 2026-10-21) and OpenAI with live model
  discovery and a platform model fallback; live evaluation with Claude and GPT
- Next.js 15 frontend with the Porcelain/Graphite design system, streaming
  chat with citations and the live Connections map
- 200 backend tests, a Playwright end-to-end suite, CI with a pgvector
  service, Docker Compose deployment, and architecture, security, testing,
  evaluation and implementation-status docs

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
main held the original CSV-AI Streamlit app (v1). This merge makes the
Knowledge Intelligence platform the content of main while keeping both
histories: the v1 commits (first parent) and the v2 -> 3.0 line (second
parent), so main fast-forwards from its previous state.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@gitguardian

gitguardian Bot commented Nov 2, 2026

Copy link
Copy Markdown

⚠️ GitGuardian has uncovered 10 secrets following the scan of your pull request.

Please consider investigating the findings and remediating the incidents. Failure to do so may lead to compromising the associated services or software components.

🔎 Detected hardcoded secrets in your pull request
GitGuardian id GitGuardian status Secret Commit Filename
39399587 Triggered Generic Password 8022182 .github/workflows/ci.yml View secret
39399587 Triggered Generic Password 8022182 .github/workflows/ci.yml View secret
39399586 Triggered Username Password 4933f00 backend/app/connectors/integrations.py View secret
39399587 Triggered Generic Password 4933f00 .github/workflows/ci.yml View secret
39399587 Triggered Generic Password 4933f00 .github/workflows/ci.yml View secret
39399586 Triggered Username Password 8022182 backend/app/connectors/integrations.py View secret
39399587 Triggered Generic Password 4933f00 .github/workflows/ci.yml View secret
39399587 Triggered Generic Password 8022182 .github/workflows/ci.yml View secret
39399587 Triggered Generic Password 8022182 .github/workflows/ci.yml View secret
39399587 Triggered Generic Password 4933f00 .github/workflows/ci.yml View secret
🛠 Guidelines to remediate hardcoded secrets
  1. Understand the implications of revoking this secret by investigating where it is used in your code.
  2. Replace and store your secrets safely. Learn here the best practices.
  3. Revoke and rotate these secrets.
  4. If possible, rewrite git history. Rewriting git history is not a trivial act. You might completely break other contributing developers' workflow and you risk accidentally deleting legitimate data.

To avoid such incidents in the future consider


🦉 GitGuardian detects secrets in your source code to help developers and security teams secure the modern development process. You are seeing this because you or someone else with access to this repository has authorized GitGuardian to scan your pull request.

Safiullah-Rahu and others added 2 commits November 2, 2026 05:47
- backend/pyproject.toml pointed readme at ../README.md, outside the
  package; current setuptools refuses that, so `pip install -e` failed in
  the backend and e2e jobs. Add backend/README.md and point at it.
- Run the disposable CI Postgres containers with trust authentication so
  no passwords appear in ci.yml (flagged by GitGuardian).
- Reword the IMAP catalog label that tripped a username/password detector.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The connector tests reached the "customer" database on the same server and
port as the platform database, which the SSRF guard deliberately refuses
(6 backend failures). Give the customer database its own Postgres service
on port 5433, as the embedded setup does locally with a second server.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@Safiullah-Rahu
Safiullah-Rahu merged commit 0cf3eed into main Nov 2, 2026
3 of 4 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant