📁Chronicle — chandlerOS
📂 Source

Chronicle

Chronicle is a local-first development memory system that records AI coding sessions, links them to git commits, and makes everything searchable through a Model Context Protocol server. 140 tests, multi-provider AI summarization, and database-tracked project management.

PythonSQLiteMCPFastMCPClaude CodeGemini
Overview

Modern developers use multiple AI coding assistants, but each session starts from scratch with zero memory. Chronicle solves this by creating a searchable, AI-powered memory system — recording full session transcripts, linking them to git commits, and exposing everything through a Model Context Protocol server that AI assistants can query directly.

Key Features
  • Full session recording across Claude Code, Gemini CLI, and Qwen with transcript capture
  • MCP server with 21 queryable tools — AI assistants can search past sessions directly
  • AI-powered summarization via Gemini and Ollama with intelligent chunking for large sessions
  • Automatic git commit linking — connects commits to the AI sessions that created them
  • Database-tracked project management with milestones, next steps, and roadmap views
  • Multi-project organization with automatic repository detection
  • Claude Skills marketplace integration with 3 pre-built skills
  • Obsidian vault export for knowledge graph visualization
  • Full-text search with boolean operators across all sessions and commits
  • Cross-platform agent prompts adaptable to Cursor, Windsurf, and other AI tools
Tech Stack
Core
  • Python 3.11+
  • SQLite
  • SQLAlchemy ORM
  • Click CLI
MCP Server
  • FastMCP framework
  • 21 tools (query + CRUD)
  • Model Context Protocol
AI Integration
  • Google Gemini API
  • Ollama (local LLM)
  • Multi-provider summarization
Testing
  • pytest
  • 140 tests across 8 modules
  • Full CRUD coverage
Challenges & Solutions
MCP Server Design

Built a 21-tool MCP server that exposes session queries, project management CRUD, and roadmap views — enabling any MCP-compatible AI assistant to query the Chronicle database directly during conversations.

Large Session Summarization

AI coding sessions can produce massive transcripts. Implemented intelligent chunked summarization that processes sessions in segments, then consolidates into coherent summaries — working across both cloud (Gemini) and local (Ollama) providers.

Cross-Platform Memory

Designed the system to work across Claude Code, Gemini CLI, and Qwen — recording sessions from any tool and making the combined history searchable from any other tool via MCP.

Outcomes
  • Open-source project that attracted recruiter attention on GitHub
  • 140 passing tests across 8 test modules with comprehensive CRUD coverage
  • Demonstrated 2,700x ROI from real-world usage — past session search saving hours of re-explanation
  • Published to Claude Skills marketplace with 3 pre-built workflow skills
  • Pioneered the concept of AI session recording as a development tool
Chronicle — 6 technologies