v1.3.0 Release Notes

What's Changed

• feat: add dbt core projects via source code parsing by @pragunbhutani in #72

Full Changelog: v1.2.0...v1.3.0

v1.2.0 Release Notes

Released: July 16, 2025*

🚀 Major Features

Model Context Protocol (MCP) Server Integration

• New Feature: Complete MCP server implementation for Claude.ai and other MCP clients
• OAuth 2.0 Flow: Secure authentication with PKCE support and organization-scoped access
• Self-hosted Only: 1:1 client-server relationship for self-hosted deployments
• MCP Tools Available:
  • list_dbt_models - Browse and filter dbt models
  • search_dbt_models - Semantic search capabilities
  • get_model_details - Detailed model information with SQL and lineage
  • get_project_summary - Project overview and structure
• Docker Integration: MCP server runs on port 8080 within Docker Compose stack
• Security Features: JWT tokens, automatic expiry, organization isolation

Landing Page Overhaul

• Complete Redesign: Modern, responsive landing page with improved UX
• New Components:
  • Enhanced hero section with better messaging
  • Bento-style features grid
  • Pain points and solution sections
  • Installation guide component
  • Updated testimonials section
  • Professional navigation bar
  • Improved footer with better organization
• Visual Improvements: Updated colors, typography, and spacing throughout

✨ New Features

Waitlist Modal

• Sign-up Flow: Users can now join a waitlist for early access
• Integration: Seamlessly integrated across landing page components
• OAuth Continuation: Enhanced OAuth flow with waitlist support

Pricing Updates

• Streamlined Pricing: Simplified pricing structure and presentation
• Color Scheme: Updated pricing section with improved visual hierarchy
• Responsive Design: Better mobile and desktop experience

📖 Documentation

Comprehensive MCP Documentation

• Setup Guide: Step-by-step MCP server configuration
• OAuth Flow: Detailed explanation of authentication process
• Security Features: Documentation of PKCE, JWT tokens, and organization scoping
• Claude.ai Integration: Instructions for connecting Claude.ai to MCP server
• Troubleshooting: Common issues and debug commands
• Production Deployment: Guidelines for production use

README Enhancements

• MCP Section: Extensive documentation (Section 8) covering all MCP functionality
• Architecture Diagrams: Sequence diagrams for OAuth flow understanding
• Example Interactions: Real-world usage examples with Claude.ai
• Security Considerations: Detailed security features and limitations

🔧 Technical Changes

Backend Changes

• MCP Server App: New Django app for MCP server functionality
• OAuth Models: New models for OAuth clients and authorization requests
• ASGI Updates: Enhanced ASGI configuration for MCP server support
• Admin Interface: Enhanced admin for managing MCP OAuth clients
• Management Commands: New command for OAuth cleanup
• Database Migrations: New migrations for MCP-related models

Frontend Changes

• OAuth Continuation: New page for handling OAuth redirects
• Landing Components: 8 new landing page components
• Waitlist Integration: Modal component for waitlist signups
• Authentication Hooks: Enhanced useAuth hook for waitlist functionality

Infrastructure

• Docker Compose: Added MCP server container configuration
• Port Configuration: MCP server exposed on port 8080
• Environment Variables: New MCP-specific environment variables
• Health Checks: Health endpoint for MCP server monitoring

🗂️ File Changes Summary

Added Files

• MCP Server: Complete new mcp_server/ directory with FastMCP implementation
• OAuth Server: backend_django/apps/mcp_server/oauth_server.py (674 lines)
• Landing Components: 8 new React components for landing page
• Waitlist Modal: frontend_nextjs/src/components/waitlist-modal.tsx
• OAuth Continue: frontend_nextjs/src/app/(auth)/oauth-continue/page.tsx
• Management Commands: OAuth cleanup command

Modified Files

• README.md: +220 lines of MCP documentation
• Integration Models: Enhanced with MCP OAuth support
• Landing Page: Complete restructure with modular components
• Authentication: Enhanced OAuth flow and session management
• Docker Configuration: MCP server integration

🛠️ Developer Experience

New Development Tools

• MCP Testing: Test suite for MCP server functionality
• OAuth Management: Admin interface for OAuth client management
• Debug Commands: Health checks and OAuth metadata endpoints

Enhanced Documentation

• Architecture: Clear separation of concerns with MCP server
• Security: Comprehensive security model documentation
• Deployment: Production-ready configuration examples

🔒 Security Enhancements

• PKCE Implementation: Proof Key for Code Exchange for OAuth security
• Organization Scoping: Automatic data isolation per organization
• JWT Tokens: Secure, stateless authentication
• Token Expiry: Automatic cleanup of expired tokens
• CORS Configuration: Proper cross-origin resource sharing setup

📊 Statistics

• Total Files Changed: 42 files
• Lines Added: ~4,000+ lines
• New Components: 8 React components
• New Models: 3 Django models
• New Features: 4 MCP tools
• Documentation: 220+ lines of new documentation

🚨 Breaking Changes

• Self-hosted Only: MCP server feature requires self-hosted deployment
• Port 8080: New port requirement for MCP server
• Environment Variables: New MCP-specific configuration required

📝 Notes

• This is a beta release with active development ongoing
• MCP server requires self-hosted deployment due to 1:1 client-server relationship
• OAuth flow is production-ready with comprehensive security measures
• Landing page redesign improves user onboarding experience
• Documentation significantly expanded for better developer experience

🔗 Related Issues

• MCP server functionality addresses enterprise integration needs
• Landing page updates improve user acquisition and onboarding
• OAuth implementation enables secure third-party integrations
• Waitlist modal supports product launch and user engagement strategies

---

Full Changelog: main...staging

🚀 v1.1.0 Release Notes

✨ New Functionality

• Model Detail View

  • /dashboard/knowledge-base/models/[id] now shows model metadata, description, and lineage.

• Per-Organisation LLM Defaults

  • Organisations can now set a default chat & completion model; falls back gracefully if unspecified.

• Secret Management v2

  • Secrets (API keys, tokens) are no longer stored raw in the database.
  • New “parameter-path” strategy backed by your secrets manager (e.g. AWS SSM).

🛠️ Improvements

• Slack Integration

  • Fixed verification URL handling when events come from multiple workspaces.

• Chat UX

  • Conversations table now surfaces model, latency & token stats.
  • “Answering…” indicators and richer logging for streaming replies.

• Knowledge-Base Table

  • Columns re-ordered, column header tooltips, new loading skeletons.

• LLM Workflows

  • Cleaner “query-verifier” tool plumbing; better SQL explanation prompts.
  • Subtle personality tweaks to reduce over-confidence.

• Slack Block Kit refinements

  • Clearer, more concise message layouts sent back to Slack.

• Subtle “personality” tuning

  • Responses feel more natural and consistent.

• Docs & Examples

  • Root README.md streamlined; .env.example updated with new vars.

🐛 Fixes

• Missing env var when bulk-toggling models → 500 (now 400 with message).
• slack_responder handler sent 404 on first install – corrected route.
• dbt project “Add” button not loading modal in some browsers.
• Secret saving race condition when updating multiple integrations.

🧩 Codebase & Infrastructure

Backend

1. New app-level signals for organisation settings defaulting.
2. Added prompt fragments to workflows/prompts/common.py.

Frontend

1. Removed deprecated integrations/mcp page.
2. integration-config-modal now supports dynamic fields & validation per provider.
3. Auth pages cleaned up (consistent error states, tailwind classes).

DevOps

• docker-compose.yml mounts new staticfiles/ volume.
• Front- and back-end Dockerfiles now use multi-stage builds for slimmer images.

🔧 Commits Overview

c06ca20 fixed slack verification url
09b8d67 added model detail display page
584ec87 improved chat metadata display
40ac8b1 fixed tool usage for query verifier
169a766 subtle personality improvements
75a742e improved answering and logging
6f61c0b minor readme tweaks
e770c72 updated readme
78bf511 added LLM defaults for orgs
0b1e9c9 fixed missing environment variable in bulk toggle
5636914 merge: sprint-to-production
76c67a2 removed need to add LLM API keys
8f600f3 fixed secret saving for api keys
bf62f89 improvements to interpretation
c134ce2 fixed dbt project addition button
362c6e5 added safer secret handling
5d64e03 improved block kit responses
12c0eda better context pulling for past conversations

---

🙏 Thank you to everyone who contributed to this tranche of work!

🔗 Full Changelog: github.com/pragunbhutani/dbt-llm-agent/compare/v1.0.0...v1.1.0

Version 1.0.0

🚀 Release Notes – fix/better-query-debugging

Merge target: main · Date: 2025-06-26

Below is a curated summary of everything that landed on fix/better-query-debugging since it diverged from main.
Thanks to everyone who contributed fixes, features and polish! 🎉

---

✨ New Features

• Monorepo Refactor – Migrated to a structured monorepo (backend_django + frontend_nextjs) with shared root tooling and Docker-Compose orchestration. (9b38309)
• Next.js App Shell & Auth Flow – Introduced a reusable AppShell layout plus fully-functional Sign-in / Sign-up pages wired to DRF + JWT via next-auth/react. (a930e0e)
• Conversations Dashboard – Added /dashboard/chats with paginated table, loading states, and suspense wrapper for smooth UX. (6daab8d, 93f6b7f)
• Pluggable Integrations System – Modularised integrations and shipped first-class MCP Integration alongside Slack, Snowflake & Metabase placeholders. (6264506)
• Knowledge Workflows Enhancements – Faster model interpretation and more robust question-answering pipeline with verified SQL fall-back. (b7cc180, fe7c30b)

🛠 Improvements

• SQL Verification – Multiple iterations improving accuracy, speed, and graceful degradation when verification fails. (7e1227a, a8dd4c2, bf3f59b)
• Query Debugging – Cleaner debug output, duplicate query removal, and concise formatting for unverified responses. (a852065, c454f1a, 65d693c)
• Auth & Security – Hardened auth checks, fixed CSRF/CORS edge-cases, and ensured token propagation to SWR fetchers. (bd3885d, bc92311, 6ee4d67)
• UX Polish – Suspense wrappers, improved loading skeletons, and better error boundaries across dashboard pages. (93f6b7f)
• Containerisation – Updated Dockerfiles & dev-containers; background job runner wired via Celery and uv Poetry replacement. (c066028, 3a571b6)

🐛 Bug Fixes

• Broken migrations & dependency ordering (4b8d52e, 273bac3)
• Next.js module resolution issues on fresh installs (4dc601c)
• Signup POST mis-routing & CSRF token mis-handling (a4fbe27, bc92311)
• Conversation logging inconsistencies and missing SQL verification links (bf3f59b)
• Misc CORS errors in local & containerised environments (6ee4d67)

🏗 Internal / DevX

• Added shared lib/ utilities for fetch wrappers & token helpers (993f1aa)
• Adopted uv for Python env management and pnpm workspaces for JS (3a571b6)
• Streamlined CI image build and multi-stage Docker workflows (c066028)

---

⚙️ Upgrade Notes

1. Database Migrations
   Run from repo root:

   uv run python backend_django/manage.py migrate

2. Docker Users
   Re-build images to pick up the new multi-stage configs:

   docker compose build --pull
   docker compose up -d

3. Local Dev
   Ensure uv (≥ 0.1.40) and pnpm (≥ 9) are installed, then:

   pnpm install         # installs JS workspaces
   uv pip install -r backend_django/pyproject.toml

---

🔍 Commits Included

a852065, c454f1a, 65d693c, f578c19, a8dd4c2, bd3885d, 93f6b7f,
bf3f59b, a4fbe27, bc92311, 6ee4d67, 993f1aa, 4dc601c, 3a571b6,
273bac3, 4b8d52e, 6264506, 6daab8d, fe7c30b, b7cc180, c066028,
a930e0e, 9b38309, 7e1227a

---

Happy shipping! 🚢

v0.9.1 Release Notes

What's Changed

• added metabase integration and the interpret and embed command (be32447)
• added the query verification flow to our question answerer (5dcbfa9)
• basic learning flow is working (d7fa20b)

Full Changelog: v0.9.0...v0.9.1

v0.9.0

What's Changed

• fixed: broken method reference for interpretation by @pragunbhutani in #60
• feat: convert app to django by @pragunbhutani in #61
• feat: added query executor to return query results to slack

Full Changelog: v0.8.4...v0.9.0

v0.8.4 Release Notes

Fixes

• fix: broken database interactions because of migrations by @pragunbhutani in #58
• fixed: serialization of question by question answerer by @pragunbhutani in #59

Full Changelog: v0.8.3...v0.8.4

v0.8.3 Release Notes

✨ Features

• Added token counting and feedback registration capabilities (e0075bd)
• Introduced provision for additional rules (817efc4)
• Implemented faster past context retrieval (91f1911)
• Enhanced context searching for improved question answering (f509a75)
• Improved answer generation and added log saving (007cabf)

🐛 Fixes

• Corrected the model list used for verification (5e0b04d)

♻️ Refactoring

• Refactored prompt structures (a6bb763)
• Moved verification and formatting logic to the Slack responder (37f083a)

New Contributors

• @HemabhKamboj made their first contribution in #56

Full Changelog: v0.8.2...v0.8.3

v0.8.2 - Politely Acknowledge

What's Changed

• feat: the bot acknowledges questions before answering them by @pragunbhutani in #55

Full Changelog: v0.8.1...v0.8.2

v0.8.1 - Faster Interpretation

What's Changed

• feat: added the ability to interpret via workflow for speed by @pragunbhutani in #54

Full Changelog: v0.8.0...v0.8.1