commit-roast
Skill14 starsChanged 6 months agoMITView on GitHub
A savage but accurate code review of your git commit history. Identifies real problems through the lens of a brutally honest senior engineer. Reads your recent commits and delivers a roast that is 30% comedy, 70% real feedback: gentle — Disappointed professor energy. "I expected better." medium — Exasperated senior dev. "We need to talk about your commits." savage — Senior engineer who has seen too much. No mercy, still fair. corporate — Passive-aggressive performance review language. "Opportunities for growth."
In this repository14 other files it gives its agents.
- llms.txtllms.txt
- claude-cost-trackerSkill
- context-budgetSkill
- debug-trailSkill
- env-doctorSkill
- ev-calculatorSkill
- git-standupSkill
- growth-trackerSkill
- investment-briefing-agentSkill
- multi-agent-orchestratorSkill
- news-sentiment-engineSkill
- options-flow-analyzerSkill
- price-monitor-alertSkill
- pr-review-prepSkill
The file
We split this skill into the rules it gives an agent
1 rule. About 0 are common, 1 is distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈1
Distinctive · estimate
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 84
No API keys needed — runs entirely on git history
≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Commit Roast
A savage but accurate code review of your git commit history. Identifies real problems through the lens of a brutally honest senior engineer.
What it does
Reads your recent commits and delivers a roast that is 30% comedy, 70% real feedback:
- Calls out vague messages ("fix", "update", "wip", "stuff")
- Flags commit size issues (300-file commits, or 47 commits to change one line)
- Spots patterns: late-night commits, "fix previous fix" chains, whitespace-only changes
- Identifies the commit where you clearly gave up (usually commit 23 of 25 on a Friday)
- Ends with 3 genuine suggestions to improve your commit hygiene
Example Output
Commit Roast — last 30 commits
THE HIGHLIGHTS:
"fix" — May 9, 2:47 AM
A 47-file commit at 3 AM called "fix." Fix what? Fix everything?
Fix your sleep schedule? We may never know.
"wip" (x4) — May 10
Four consecutive WIP commits. You committed "wip" to main.
To the main branch. Four times.
This is not a draft. This is a published novel.
"update utils" — May 11
Updated utils.ts. All 847 lines of it.
One commit. 847 lines.
Reviewers saw this and went home early.
"fix previous fix" — May 12
The ouroboros commit. The fix that ate its own tail.
The commit that proves the previous commit should not have existed.
WHAT YOU DID WELL:
- Commits on May 8 were actually good. Specific, scoped, readable.
"feat(auth): add refresh token rotation with 7-day expiry" — chef's kiss.
- You rebased before the PR. Respect.
GENUINE FEEDBACK:
1. Your commit messages average 4 words. Aim for 8-12. Subject + brief why.
2. Three "wip" commits = one real commit. Squash before pushing.
3. 2 AM commits average 3x more "fix" commits the next morning. Sleep first.
VERDICT: B- (was A- before the 3 AM incident)Usage
Roast my last 30 commits.
Be honest about what is bad, but make it funny.
End with 3 real actionable suggestions.
Severity: medium (not mean, not toothless)Severity modes
gentle — Disappointed professor energy. "I expected better." medium — Exasperated senior dev. "We need to talk about your commits." savage — Senior engineer who has seen too much. No mercy, still fair. corporate — Passive-aggressive performance review language. "Opportunities for growth."
What it actually catches
Beyond the jokes, this skill identifies real commit hygiene problems:
| Pattern | What it means |
| "fix" / "update" / "wip" | No searchable history — future you will hate past you |
| 100+ file commits | Unreviewed blast radius — something will break |
| Fix-the-fix chains | Original design was wrong and got shipped anyway |
| Late night commit spikes | Rushed work that often needs follow-up |
| Identical message repeated | Copy-paste commit workflow |
Requirements
- Bash tool access (reads
git log) - No API keys needed — runs entirely on git history
Free
Fully free and open source. Part of the AI Investment Skills Bundle (the bundle is serious; this skill is for fun).
→ https://github.com/tellmefrankie/news-engine
Author
Built after reviewing a PR with 47 commits where commit #23 was literally "asd". The roast practically wrote itself.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

