SlopScore
00 crowd

ai-commit-sweeper

Find and prevent AI-agent attribution artifacts (session URLs, Generated with Claude Code, AI co-author trailers) in git commits
Open repo on GitHubgithub.com/w1977-0/ai-commit-sweeper
Python · ★ 1 · 0 forks · MIT · paperwork by the Cap'mmostly ai (inferred)light human (inferred)works-on-my-machine (inferred)clidevtools🤖 aider🤖 claude-code🤖 copilot🤖 cursor
listed 2 hours ago by w1977-0 · last checked 16 minutes ago
The owner didn't write this. This repo never submitted itself. The Cap'm found it on a truffle trawl and wrote its paperwork from what GitHub already shows. Picked by hand by the Cap'm on 2026-09-28: Find and prevent AI-agent attribution artifacts (session URLs, Generated with Claude Code, AI co-author traile; its own README says "Find and prevent AI-agent attribution artifacts (session URLs, Generated with Claude Code, AI co-author trailers) in git commits". 1 stars; MIT license. The owner did not submit this. Votes count; awards don't until the owner claims it.

I'm not calling your project slop! Geeze, it's a joke... Do you own this repo?

Log in with GitHub as w1977-0. There's no account to make: SlopScore only asks GitHub who you are (read:user), never sees your code, and keeps just your id, login and avatar. Then you can:

  • Keep it, on your terms. Commit your own slopscore.md (spec) and press Refresh. Your paperwork replaces the Cap'm's, and you can submit it for Slop of the Day.
  • Take it down. One click on Remove. It stays gone; the trawl never brings it back.

Log in with GitHub

Can't log in as the owner? Request a takedown. No login needed, and a trawled listing comes down right away.

GitHub says
Find and prevent AI-agent attribution artifacts (session URLs, Generated with Claude Code, AI co-author trailers) in git commits
topics
aiaiderclaude-codeclicommit-hygienecopilotcursordeveloper-toolsgitpython
created
2026-08-31 · pushed 5 days ago · 7 commits · 1 contributor
languages
Python 100%
paperwork
code of conductcode of conduct filecontributingpull request templatelicensereadme 100% health
dependencies
no dependency graph (no manifest, or disabled) · OSV.dev, checked 2 hours ago

Disclosures, inferred by the Cap'm

slopbucket
vibe-coded
category
clidevtools
ai_generated
mostly
human_touch
light
status
works-on-my-machine
built_with
aiderclaude-codecopilotcursor
language (detected)
python
topic (detected)
aiaiderclaude-codeclicommit-hygienecopilotcursordeveloper-toolsgitpython
license (detected)
mit

The Cap'm's log

The Cap'm wrote this paperwork, not the owner. This repo never submitted itself to SlopScore. The Cap'm picked it by hand: Find and prevent AI-agent attribution artifacts (session URLs, Generated with Claude Code, AI co-author traile; its own README says "Find and prevent AI-agent attribution artifacts (session URLs, Generated with Claude Code, AI co-author trailers) in git commits". It carries the MIT license. The disclosures above are his best guess from what GitHub shows.

Is this yours? Commit a real slopscore.md and press Refresh to replace this, or remove the listing in one click. There's no account to make: you log in with GitHub.

README — the repo's own words, folded up so the grading fits on one screen

aisweep

AI attribution artifacts, swept out of your git history.

Find and block the session URLs, Generated with Claude Code lines and AI co-author trailers that AI coding agents leave in your commits — for developers who use Claude Code, Cursor, Copilot, Aider, Gemini CLI or Codex and publish clean repositories.

CI License: MIT Python Code style: ruff Tag Stars

English | 简体中文

aisweep scanning a repository and finding AI attribution artifacts

Why this exists

AI coding agents sign your commits by default, whether you asked for it or not. Claude Code appends a session URL (https://claude.ai/code/session_…) plus a Generated with Claude Code line and a Co-Authored-By: Claude trailer to every commit and PR description it writes — silently, opt-out, discoverable only after it is already in your history (anthropics/claude-code#66504, 89 👍 as of August 2026, and a lively HN thread).

Claude Code now has a setting to turn this off per tool. That still leaves:

  1. history already written — the artifacts stay in every past commit;
  2. every other tool — Cursor, Copilot, Aider & friends each have their own behaviour and their own switch;
  3. teammates — whose ~/.claude/settings.json you do not control.

aisweep is the one tool-agnostic gate: audit what is already in your history, and block polluted commits before they land.

Features

🔍 Audit scan full history (or any range) for AI artifacts, per-commit report
🛡️ Prevent a commit-msg hook stops polluted commits before they land
🧰 Tool-agnostic pattern-based — works with any agent, including future ones
🧩 Configurable .aisweep.json: disable rules, tune severity, allowlist, add your own
📤 Reports Markdown + JSON output for CI, PR review or pre-open-sourcing checks
🪶 Zero dependencies Python ≥ 3.10 standard library only; pip install and go
🤝 Non-destructive never rewrites, amends or touches a single commit

Quick start

Prerequisites: Python 3.10+ and git.

# install
pip install git+https://github.com/w1977-0/ai-commit-sweeper.git
# or isolated:  pipx install git+https://github.com/w1977-0/ai-commit-sweeper.git
# or without installing:
git clone https://github.com/w1977-0/ai-commit-sweeper.git
cd ai-commit-sweeper
python -m aisweep --help
# 1. audit a repository
aisweep scan ./my-project

# 2. prevent future artifacts
cd ./my-project
aisweep hook install

Usage

Audit existing history

$ aisweep scan ./my-project
aisweep 0.1.0 — scanning main in ./my-project
  a4e8601 2026-08-29 ✖ block claude-session-url     https://claude.ai/code/session_a1b2c3d4e5f6
  a4e8601 2026-08-29 ✖ block claude-generated-with  🤖 Generated with [Claude Code](https://claude.com/claude-code)
  a4e8601 2026-08-29 ✖ block claude-coauthored      Co-Authored-By: Claude <noreply@anthropic.com>
  24a220d 2026-08-26 ⚠ warn  ai-coauthored          Co-Authored-By: Cursor <agent@cursor.sh>
2/12 commit(s) contain AI attribution artifacts (4 finding(s)).
Prevent future artifacts in this repo: aisweep hook install

Exit code is 1 when artifacts are found, so it drops into CI and scripts directly. Useful flags: --all (scan all refs), --rev, --since, --max N, --json, -o REPORT.md.

Block polluted commits before they land

$ aisweep hook install
commit-msg hook installed at .git/hooks/commit-msg (default (block-level rules only))

$ git commit --allow-empty -m "fix: quick patch" -m "https://claude.ai/code/session_zz9x8y7w6v"
aisweep: commit message contains AI attribution artifacts:
  ✖ block claude-session-url     https://claude.ai/code/session_zz9x8y7w6v
commit blocked. Edit the message (drop the artifacts or configure .aisweep.json), then commit again.

$ git commit --allow-empty -m "fix: quick patch"   # clean messages pass

The hook enforces your policy: --strict also blocks warn-level rules, and .aisweep.json can flip any rule either way.

Shareable reports

$ aisweep report ./my-project -o REPORT.md --json-output REPORT.json
report: 4 finding(s) across 12 commit(s) -> REPORT.md

Markdown for humans, JSON for machines — handy before open-sourcing a repo that was built with AI assistance.

Default rules

Rule Severity Catches
claude-session-url block session URLs like https://claude.ai/code/session_… (#66504)
claude-generated-with block the Generated with Claude Code attribution line
claude-coauthored block Co-Authored-By: Claude <noreply@anthropic.com> trailers
ai-coauthored warn Co-Authored-By trailers naming other agents (Cursor, Copilot, Aider, Gemini, Codex, …)
ai-generated-with warn Generated with <tool> lines for other agents

Matching is case-insensitive and line-anchored, so ordinary prose (a link to the docs, a colleague named "Claude" in a normal co-author line) is not swept up by accident. See configuration for the escape hatches.

Configuration

aisweep init writes a starter .aisweep.json at the repository root:

{
  "disable": ["ai-coauthored"],
  "severities": {"ai-coauthored": "block"},
  "allow_patterns": ["^co-authored-by:.*\\bmy-team-bot\\b"],
  "extra_rules": [
    {
      "id": "acme-deploy-bot",
      "description": "our deploy bot trailer",
      "pattern": "^deployed-by: acme-bot",
      "severity": "warn"
    }
  ]
}
  • disable — switch built-in rules off (e.g. your team wants attribution).
  • severities — per-rule block / warn override.
  • allow_patterns — lines matching these regexes never produce findings.
  • extra_rules — your own patterns, compiled like the built-ins (case-insensitive, multiline).

How it works

flowchart LR
    A[git commit] --> B[commit-msg hook]
    B --> C{aisweep check}
    C -- clean --> D[commit lands]
    C -- artifacts --> E[commit blocked]
    F[aisweep scan] --> G[findings / MD+JSON report]
Loading

scan reads commit messages via git log and matches every line against the effective ruleset; check does the same for a single message file inside the commit-msg hook. aisweep is read-only: it never rewrites, amends or stages anything.

FAQ

Does it rewrite my history? No — that is out of scope by design. Use git-filter-repo for rewrites; aisweep gives you the precise map of what to rewrite and keeps new pollution out afterwards.

Is this anti-AI? No. The request behind #66504 is that attribution should be opt-in. aisweep enforces exactly that: your repository, your policy. If your team wants attribution, configure it in.

What about false positives? Patterns are narrow and line-anchored; anything that slips through can be allowed via allow_patterns or downgraded via severities.

Does it work on Windows? scan, check and report run anywhere Python 3.10+ runs. The hook is a POSIX shell script and works in Git Bash environments. CI tests on Linux.

Which tools does it cover? Any tool whose artifacts end up in commit messages — the rules match shapes, not vendors, and extra_rules covers anything we missed.

Roadmap

  • Publish to PyPI
  • GitHub Action wrapper (scan a repo in CI without installing)
  • HTML report with filtering
  • PR-description audit via GitHub API
  • pre-commit framework integration

Contributing

Issues, bug reports and PRs are welcome — see CONTRIBUTING.md for the development setup, commit style and test requirements. Please note the Code of Conduct and the security policy.

License

MIT © 2026 w1977-0

Acknowledgments

Read the rest on GitHub

Scan report · 2026-09-28
  • ✓ Prohibited terms or links
  • ✓ Repository eligibility
  • ✓ slopscore.md paperwork
  • ✓ Content policy
  • ✓ Risk review — +10 owner has 0 followers

0 comments

log in to comment.

report this listing — log in to report