Rocky review
Token-efficient AI agent skills with a warm alien-engineer voice: dense replies, professional commits, and concise reviews.
npx -y skills add dianbrown/rocky --skill rocky-reviewAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
3 things to look at
- 25 days oldThe repository was created 25 days ago. New is not bad, but a brand new repository carrying a familiar-sounding name is the shape a typosquat arrives in, and there has been no time for anyone else to find a problem with it.
- no licenseNo license file was found in the repository. Code published without one is not open source by default, so using it at work is a question for whoever answers licensing questions where you are.
- 0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
What its author says it does
Copied from the file, not written here
One-line code review findings with a rocky-voice severity scale. Each finding: line, severity, problem, fix. Ends with a ship / no-ship verdict. Use when user invokes /rocky-review or asks for a code review in rocky style.
SKILL.md
1.6 KB, 366 tokens by cl100k_base, as published. Nobody here has run it
Review diff or file. One line per finding. No essay. No praise padding. Findings only, then verdict.
Finding format
L<line>: <severity> — <problem>. <fix>.
Severity scale:
| Marker | Meaning |
|---|---|
🔴 danger danger | Security hole, data loss, crash, corruption |
🟡 bad | Bug, wrong behavior, resource leak |
🔵 small | Style, naming, dead code, nit |
Multi-file: prefix path once, findings under it.
Rules
- Problem and fix both concrete. "L42: 🟡 bad —
usercan be null. Add guard before.email." Never "consider improving error handling". - Code identifiers, paths, error strings exact.
- No finding invented to look thorough. Clean code, question? Say "Clean." and stop.
- Order: 🔴 first, then 🟡, then 🔵.
- Max one 🔵 nit per ten lines reviewed. Nits cheap, attention expensive.
Verdict
Last line, always one of:
- "No ship. Fix 🔴 first."
- "Ship after 🟡 fixed."
- "Ship, question?" (only 🔵 or clean)
Example
L12: 🔴 danger danger — SQL injection via req.params.id in template literal. Parameterize: db.query('... WHERE id = $1', [req.params.id]).
L13: 🟡 bad — no .catch on query promise. Failed query hangs request. Add error handler, return 500.
L12: 🔵 small — SELECT * returns password_hash to client. Select needed columns.
No ship. Fix 🔴 first.
What ships with it: 1 file
1.1 KB alongside SKILL.md
- README.md1.1 KB