Graft Hits 1.6k Stars: Open-Source Context Layer for Coding Agents
shhdwi · reddit · 2026-08-12
Open-source project Graft announced it has reached 1,600 GitHub stars. The tool aims to solve the common pain point of AI coding agents re-exploring the entire codebase from scratch during every session.
Core Features & Mechanism:
- Persistent Codebase Map: Maps the codebase structure into linked markdown files committed to Git, allowing agents to read the map instead of starting from zero.
- Tree-sitter Powered: The structural layer uses tree-sitter, requiring no LLM and working on any repository size.
- Provider-Agnostic: Supports direct API calls (like OpenAI) for optional LLM-generated summaries, avoiding model lock-in.
- Seamless Integration: Integrates with tools like Codex via MCP and an AGENTS.md file, featuring a post-edit hook to keep the graph synced automatically.
Benchmark Results:
In a 162-task benchmark, Graft reduced tool calls by 46%, tokens by 42%, and execution time by 60%, while maintaining steady correctness.
More from coding & agent
- RAG Trap: Vector Search Can Find the Answer Yet Fail the LLM — WirelessLife · 2026-08-12
- GitHub Repo Tracks the Evolution and Frameworks of Self-Evolving AI Agents — tom_doerr · 2026-08-12
- Developer Releases Caffeine: Keeps Mac Awake While Claude Code and Codex Run — iannuttall · 2026-08-12
- Inkbox Launches Agent2Agent Platform for Direct Cross-System Agent Communication — ycombinator · 2026-08-12
- The Bitter Lesson of Agents: Minimal Toolsets Outperform Complex Custom Harnesses — pcuenq · 2026-08-12
- One Prompt to 3D Game: Thrixel and Claude Code Workflow Open-Sourced — RanaHanocka · 2026-08-12