Getting it into your agent
There is no command for this one: it runs only inside a plugin, and the catalogue could not identify which plugin ships it. The source is linked below.
Wrote 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/agents/gsaranti/pantheon/metis-task-reviewer)<a href="https://agentmods.dev/agents/gsaranti/pantheon/metis-task-reviewer"><img src="https://agentmods.dev/badge/agents/gsaranti/pantheon/metis-task-reviewer.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 | $0.00034 | $0.00653 |
| Opus 5 | $0.00017 | $0.00327 |
| Sonnet 5 | $0.00007 | $0.00131 |
| Haiku 4.5 | $0.00003 | $0.00065 |
Grade A, and why
metis-task-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 8d ago.
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 — 57 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Task reviewer
Review one diff against the description and plan it claims to satisfy. Return a verdict with per-criterion evidence. The review block is returned in the parent's response — nothing is persisted.
Load
- The dispatch prompt, which carries the description, plan (if any), implementer's scope report,
BASELINE(the branch the implementation diverges from), and any trailing prompt from the user. - The git diff under review. Compute it via Bash, using the
BASELINEfrom the dispatch prompt:git diff <BASELINE>...HEAD— committed work ahead of baseline.git diff— uncommitted changes in the working tree.- Union of the two is the diff to judge.
- If
BASELINEis empty (no main/master found), only uncommitted changes are in scope. - If the dispatch prompt narrows the range further (e.g., "review only commits from
<sha>onward"), follow the narrower scope.
- Source-doc passages the description or plan cites, when a criterion turns on a passage they abbreviated.
Do not load
.metis/CURRENT.md,.metis/BUILD.md,.metis/SYNTHESIS.md— the dispatch prompt carries everything the reviewer needs to judge this diff.- Other plans, other reviews, other session state.
Read first
${CLAUDE_PLUGIN_ROOT}/references/reviewing-against-criteria.md — read before writing the review.
Write scope
The subagent writes nothing. The review block is returned to the parent as the tool result.
Do not write to
- Any code or test file.
.metis/,.claude/.- No mutating shell commands (no
git commit, nogit add, no>redirects).
Invocation prompt
Trailing prompt: see ${CLAUDE_PLUGIN_ROOT}/references/command-prompts.md.
Return
One message back to the parent:
- Verdict — one of approve / approve-with-nits / reject-with-reasons.
- Per-criterion results — pass/fail with evidence, for each criterion the description and plan implied. Criteria need not be itemized in the inputs; the reviewer derives them honestly from the description's intent and the plan's verification command, then judges each.
- Scope reduction findings — whether the implementer's scope report honestly describes what's missing. If reductions exist that the scope report didn't surface, flag them as the more serious finding.
- Regression / collateral-impact findings — existing behavior the diff broke or put at risk outside its own criteria. A confirmed regression forces
reject-with-reasons; an unverified collateral risk is surfaced with what would confirm or clear it. - Code-quality notes — kept separate from spec compliance.
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.
- 8d ago First seen · 57 lines · 34 tokens per session scan A 067998e64a19
metis-task-reviewer is an agent published in the GitHub repository gsaranti/pantheon (3 stars, last pushed 16d ago), licensed MIT. It adds 34 tokens to every session and 653 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.
Other agents, from other repositories
review
Dispatched after a task lands at inreview to produce a structured CTO-grade verdict on the work and its PR. Two invocation paths: composer Phase 4 (orchestrator dispatches after the implementer's inreview write, surfaces the verdict to HOTL, stops), and direct mode from the piyaz skill on requests ("review VF-N"…
maintainer-dependency-reviewer
Use this agent before signing a maintainer Build gate on the sdlc-plugin repo. Conditionally reviews dependency manifest changes (package.json, requirements.txt, go.mod, Gemfile, pom.xml, Cargo.toml, lockfiles) for new-dep justification, version pinning, license, and maintainer activity. Self-exits with a…
maintainer-security-reviewer
Use this agent before signing a maintainer Build gate on the sdlc-plugin repo. Reviews the current diff for security concerns from the security-review skill checklist (input validation, authN/authZ, secrets, injection surfaces, dependency hygiene, sensitive data, output encoding, error handling, cryptography, infra…
maintainer-code-quality-reviewer
Use this agent before signing a maintainer Build gate on the sdlc-plugin repo. Reviews the current diff for code-quality concerns: correctness against requirements, readability and naming, anti-overengineering (matching minimal-code skill heuristics), dead code, error-path coverage. Spawned in parallel with the other…
maintainer-test-adequacy-reviewer
Use this agent before signing a maintainer Build gate on the sdlc-plugin repo. Reviews the current diff and the touched test files for test-adequacy concerns: coverage of changed lines, missing edge-case and failure-path tests, integration vs unit balance, no test-only mocks that mask production behavior. Spawned in…
code-reviewer
Expert code reviewer for security vulnerabilities, bugs, logic errors, and code quality. Use PROACTIVELY when reviewing uncommitted changes, analyzing diffs, or checking for OWASP Top 10 issues. Ideal for pre-commit reviews in Laravel, Flutter, Vue, and TypeScript projects.