Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add Rishi4792/compass --skill review-buildgit clone --depth 1 https://github.com/Rishi4792/compassWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/rishi4792/compass/review-build)<a href="https://agentmods.dev/skills/rishi4792/compass/review-build"><img src="https://agentmods.dev/badge/skills/rishi4792/compass/review-build.svg" alt="Measured on agentmods" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00095 | $0.08256 |
| Opus 5 | $0.00048 | $0.04128 |
| Sonnet 5 | $0.00019 | $0.01651 |
| Haiku 4.5 | $0.00010 | $0.00826 |
Grade A, and why
review-build scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured yesterday.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 287 lines — stays where its author put it; the contents beside it link to each section on GitHub.
compass:review-build (Review-3 · FULL)
Lens: is the BUILT thing correct, complete vs the contract, and safe — proven, not vibed?
Read these before you start. They are the standards this stage is held to, and they
live in plugins/compass/shared/ so they are the same for every stage that uses them:
shared/review-core.md— the severity bug-bar, self-refutation and dedupe rules every review shares.
(A standard nobody loads is not a standard. shared/MANIFEST declares who reads each file and
doctrine-wired-check.sh proves it — feynman.md sat unread for three releases while its own
first line claimed three stages loaded it.)
Step 0a — the cheap checks run FIRST, before any reviewer is spawned
Run bash plugins/compass/scripts/mechanical-suite.sh . and read its output before doing
anything else in this stage.
Why it comes first. The last release ran six independent adversarial reviews and they found 59 recorded defects. 32 of them — just over half — needed no judgment at all: a duplicate constant, a test whose population could be empty, a check nobody called, a shell mistake. A reasoning model was paid to find the fourth copy of a literal. The suite finds that whole class in seconds, for no tokens, and it runs before a reviewer is asked to look at anything.
What to do with it:
- A MEASURED failure blocks this stage. Fix it and re-run. No reviewer should ever be handed a defect a script had already decided.
- Its REPORTED lines are handed to the reviewers as context, not as findings — they are the things the suite could not decide, which is exactly where judgment is worth paying for.
- If a reviewer finds a defect of a class the suite could have caught, ADD THE CLASS to
plugins/compass/scripts/mechanical-suite-classes.mdrather than re-reviewing it forever. That file is the registry; a class in it is a class no future review pays for twice.
Honest limit, stated so nobody over-trusts this. Only some defect classes are soundly mechanical — building this suite found that of five shell-trap classes exactly one could be decided by a line scan without failing on correct code. Every check therefore says whether it MEASURES or REPORTS. A green suite means the cheap classes are clear; it does not mean the work is right.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- yesterday Changed · +25 lines 121dbfaf51b8
- 8d ago First seen · 262 lines · 95 tokens per session scan A b0a83a68e0a9
review-build is a skill published in the GitHub repository Rishi4792/compass (2 stars, last pushed 2d ago), licensed MIT. It adds 95 tokens to every session and 8,256 once invoked, about $0.0005 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other skills, from other repositories
implement
Implement a ticket or feature with proportional planning, test first behavior changes, independent review when available, and verification through the real affected surface.
test-audit
Audit test suites for T1-T4 violations using AST analysis, mock detection, and multi-stage synthesis. Invoke when user asks to audit tests, check test quality, find mock violations, review test effectiveness, or inspect test suites for over-mocking. Triggers automatic rewrites when quality gates fail.
g-review
Run the review gate on the current branch diff. Runs the test suite, captures the diff, and dispatches code-lead, which verifies done conditions and reviews the diff itself. Issues MERGE READY or HOLD.
g-audit
Full-codebase or targeted code quality audit. Detects SOLID violations, code smells, architectural drift, dead code, and test coverage gaps. Targeted scope produces an inline report. Whole-codebase scope produces a prioritised roadmap milestone.
g-refactor
Guided refactor workflow — identify target, pre-analyse, spec, approve, execute, review. Accepts a scope path, an audit milestone file, or runs interactively. Safe-by-default: checks test coverage before execution and runs the full review gate after.
dogfood
Fix-through loop — invoke a new thing, fix bugs until it works.