- Home
- Skills
- Agents & Orchestration
- The Lorekeeper: what your repo refuses to become
More screenshots
Works with the AI Agents you already use
The Lorekeeper: what your repo refuses to become
Mines your git history for what the team reverted, removed and declined, and writes a VISION.md of evidence-backed principles so agents stop proposing changes you would reject.
$12
The Lorekeeper: what your repo refuses to become
Example session with this skill installed
Coding agents keep proposing changes we would never merge. Write down what this repo refuses to become, as a VISION.md, based on its real history. gh is logged in, read-only.
- Read your context and instructions
- Compiled the the lorekeeper
3. Maintainability and safety beat raw speed
A large rewrite of the C code for speed is declined. The stated direction for compiled code is Rust (open PR #438). Small, clear fixes for real slowdowns are welcome.
Evidence
- #519 (faster C rewrite) closed by the maintainer, who prefers the Rust version for maintainability and safety.
- ff06d77 (#522) merged a targeted fix for a quadratic striptags slowdown.
$ python3 /tmp/vision.py check VISION.md
VISION.md: 9 principles, 5 rulings, every cited sha exists
Connects securely to your tools. The creator never sees your data.
About this skill
Remembers every time you said no.
Coding agents keep proposing changes that look fine and that your maintainers would reject on sight: a plugin system in a project that already removed one, a runtime dependency in a library that avoids them. The reasons are in your history. Nobody wrote them down.
What it does
- A small Python script reads your local git history and prints an evidence digest: reverts, directories that no longer exist, commits that mostly deleted files, commit messages that say "remove", "drop", "instead of" or "rather than", merged work and high-churn files. Every row starts with a real commit sha.
- Optionally, with your own
ghCLI, it adds PRs closed without merge (with the last comment) and the longest review threads. - The agent reads the strongest leads and drafts 5 to 9 principles. Each one says what the repo resists and cites a commit sha or PR number. No evidence, no principle, unless you state it yourself.
- It stress-tests the draft with 3 to 5 hard hypothetical changes and asks you to rule on each.
- It writes
VISION.mdwith your rulings, runs a check that every cited sha exists, and offers one pointer line for CLAUDE.md or AGENTS.md.
What you get
A short, reviewable file in your repo that tells any agent, and any new teammate, what this project will not become, and why, with links back to the moments the team decided.
Who it is for
Maintainers of open source libraries, tech leads of long-lived services, and anyone who reviews a lot of agent-written PRs.
Part of The Guild, agent skills by Sesisnando Neto: The Oathkeeper (sealed acceptance checks), The Quartermaster (quests and decisions), The Tracker (root-cause proof), The Archivist (memory hygiene), The Inspector (QA validation).
Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Agents & Orchestration.
Needs real history: it stops below 30 commits unless you force it. Commit messages are matched by keywords, so many rows are routine and the agent has to judge them. Shallow clones hide old history. PRs closed without merge are only found with gh on GitHub. The principles are the agent's reading of the evidence until a maintainer rules on them.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before buying a skill.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed today
- Accounts
GitHub
- Software
Git
Python 3.8+Bash
- Optional
- GitHub CLI (gh), logged in
Permissions
Allowed Hosts
File Scopes
Read-only on git and GitHub. Writes VISION.md in your working tree. The optional --gh step reads PR titles and comments through your own gh CLI (api.github.com); nothing is posted. Never commits, pushes, comments or reviews.