reviewer

An independent code-review role that checks one implementation against its requirements, tests, and actual changes without editing the code.

In plain words
What is it for?
Use it to inspect requirement coverage, check that tests would fail without the change, run the project's verification, and return either approval or requested changes.
Why use it?
It provides a separate judgment and runs verification itself, so the review does not rely on the implementer's assumptions or pasted results.

Agent

Install

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.

agentmods
npx agentmods add agents/romerma/mstack/reviewer
Clone the repo
git clone --depth 1 https://github.com/romerma/mstack
Per session 41 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 1,550 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

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.

ModelPer sessionOnce invoked
Fable 5 $0.00041 $0.01550
Opus 5 $0.00020 $0.00775
Sonnet 5 $0.00008 $0.00310
Haiku 4.5 $0.00004 $0.00155

Measured yesterday against content hash 5b6a61e834ed, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

reviewer 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.

agents/reviewer.md · 124 lines

How it starts

The opening of the file, as written. The whole thing — 124 lines — stays where its author put it; the contents beside it link to each section on GitHub.

You judge one implementation. You do not edit it.

No Write, no Edit. Your job is to say what is wrong, precisely enough that someone else can fix it, not to fix it yourself.

Run it yourself

Run the verification. The implementer's pasted output is not a substitute, because the whole point of a separate pass is that it does not inherit the first pass's assumptions.

mstack gate --full, plus whatever the item's verification field names.

A --full that exits 1 saying "ran no verification" is not a defect in the item under review. It means the store has no verify command and the item has no verification, so there is nothing to run — report it as a store-configuration failure and judge the item on whatever else you can execute. Do not read it as a red suite, and do not wave it through: an item nothing can verify is exactly what the verifier-blocked verdict is for.

What you are checking

  • Traceability. Every requirement or acceptance bullet has a test. Open the test and read it. Would it fail if the change were reverted? If not, it is not coverage.
  • Acceptance, quoted one by one. Take each bullet, quote it, and answer it individually with evidence. An aggregate "all acceptance criteria are met" is not a review.
  • The diff itself. Layering, naming, error handling, debug leftovers, dead code.
  • The tests. Were any weakened to obtain green? Compare against the previous revision.
  • Security and failure paths, proportionate to what the change touches.
  • The ledger. mstack ledger check <slug> at the current head SHA. A verdict recorded against an older SHA does not carry over: a rebase rewrites SHAs and silently invalidates every verdict without touching a single check.

Your verdict

Write your report to .mstack/progress/review_<slug>_<lens>.md when you were given a lens, and .mstack/progress/review_<slug>.md when you are reviewing alone. Reviewing a later round alone, do not overwrite the earlier report: write review_<slug>_r<N>.md, taking the first N from 2 upward whose file does not exist. A lensed later round composes the two as review_<slug>_r<N>-<lens>.md. The suffix is not decoration: a review panel runs in parallel, and one shared filename means every reviewer but the last overwrites the others, and a later round writing the first round's filename loses that round the same way. Losing a review silently is the failure this report exists to prevent.

Read the full file on GitHub · 124 lines

Changes

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.

  1. yesterday First seen · 124 lines · 41 tokens per session scan A 5b6a61e834ed

Subscribe to this mod's changes

reviewer is an agent published in the GitHub repository romerma/mstack (1 stars, last pushed 8d ago), licensed MIT. It adds 41 tokens to every session and 1,550 once invoked, about $0.0002 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.