2.5k Downloads
Overview
Automatically maintains a human-readable PROJECT_STATE.md after each git commit, summarizing recent changes, current focus, and suggested next steps using either raw git data or optional AI-generated summaries.
Key Advantages
1.Hands-free, always-fresh project context via git post-commit hook
2.Combines structured git metadata with AI summaries for fast mental model of the repo state
3.Configurable sections and behavior via .project-context.yml (AI on/off, recent commits, diff stats, sections)
4.Safe default behavior (falls back to non-AI mode if Clawdbot/gateway isn’t available)
5.PROJECT_STATE.md is gitignored, avoiding merge conflicts and noisy diffs in the main repo history
Use Cases
- Keeping a single, up-to-date status file for agents or future sessions to understand project state quickly
- Improving handoff between developers by providing a narrative of recent work and current focus
- Onboarding new contributors with a concise history and description of what’s currently being worked on
- Supporting AI coding agents that need lightweight, structured context without scraping git history each time
- Personal productivity for solo devs to remember what they did and what to do next after breaks or context switches
Evaluation Scores
8.5
/ 10
Reliability
8.0
Functionality
9.0
Usability
8.8
Safety
8.0
Performance
8.5
Compatibility
8.5
Based on 1 evaluation · Latest: 3/20/2026
Download Trend
Loading...
Evaluation History (1)
8.5/103/20/2026▼
OS: linux-x64LLM: google/gemini-3-flash-preview
**Judgement:** A well-scoped, practical skill for maintaining a living project status document tied to git commits. It’s especially useful in AI-assisted workflows and for developers who frequently switch contexts or share projects.
**What it’s good at:**
- Automatically generates and updates `PROJECT_STATE.md` on every commit using a post-commit hook.
- Captures last commit details, recent history, branch, changed files, and optional diff stats.
- Optionally calls an LLM (via Clawdbot’s HTTP gateway) to infer current focus and suggest next steps.
- Offers a non-AI mode that uses only raw git info, which is fast and cost-free.
- Easy installation/uninstallation scripts and a straightforward YAML config file.
**Key risks / limitations:**
- **Gateway/LLM exposure:** AI mode sends commit/context data through Clawdbot’s OpenAI-compatible endpoint. While it’s loopback-bound and token-protected by default, misconfiguration or credential leakage could expose sensitive repo information.
- **Hook fragility:** The post-commit hook can be overwritten by other tools or manual changes, breaking automatic updates unless reinstalled.
- **Latency & cost in AI mode:** Each commit may incur LLM latency and token costs, which can add up in very high-commit workflows.
- **AI reliability:** Summaries and “suggested next steps” may occasionally be inaccurate or overconfident; they should be treated as guidance, not ground truth.
**Recommended scenarios:**
- Individual developers or small teams using Clawdbot who want a continuously updated, human-readable project status file.
- Repos where AI agents or tools frequently need a quick snapshot of recent work and current focus.
- Projects with frequent handoffs or intermittent work, where remembering context and next steps is a recurring pain point.
**Less ideal for:**
- Highly sensitive codebases where sending any project context to an LLM (even via a local gateway) is disallowed.
- Workflows that already use heavily customized git hooks that might conflict with additional post-commit scripts.
- Extremely high-frequency commit patterns where additional post-commit processing time is unacceptable.
Comments (0)
No comments yet. Be the first!