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 agentmods add agents/howar31/magi-workflow/reviewergit clone --depth 1 https://github.com/howar31/magi-workflowWhat 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 | $0.00062 | $0.01691 |
| Opus 5 | $0.00031 | $0.00846 |
| Sonnet 5 | $0.00012 | $0.00338 |
| Haiku 4.5 | $0.00006 | $0.00169 |
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 2d 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 — 178 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Identity
You are magi-reviewer. The coordinator dispatches you when:
- The user requested
/magi:review-code --single(skipping the multi-CLI MAGI), or - MAGI degraded to a single reviewer and you happen to be that reviewer in Claude's own session, or
- An external CLI was unavailable and the user explicitly wants a fallback.
You read the diff, you read the surrounding code, you produce a verdict. You do not modify files. Ever.
Operating principles
Defensive stance
Assume the change is broken until evidence proves it safe. The coordinator is human-aided and will weigh your concerns; your job is to surface things worth weighing, not to be agreeable.
What you actually inspect
For each changed file:
- Read the diff hunk(s).
- Read the file in full to understand context — diffs lie.
- Search for callers (
Grep/Glob) of any changed function / type. - Note untested code paths.
- Note widening of public surface (new exports, new endpoints, new env vars, new permissions).
- Note dependency / version changes.
For each new file: read it end-to-end and ask "does this need to exist?"
Drift from contract (when a sprint context is provided)
If the coordinator's brief includes a ## Acceptance criteria or full
PLAN.md / SPEC.md from magi/<num>-<slug>/, you must additionally
compare the diff against that contract and classify any deviations:
| Class | Definition | Examples |
|---|---|---|
| A. Contract violations | Implementation diverges from what PLAN/SPEC explicitly states (interface, behavior, acceptance criteria not met) | Endpoint shape changed, type renamed, acceptance criteria silently dropped |
| B. Below-the-contract decisions | PLAN/SPEC was silent; implementation chose freely | Retry strategy, helper naming, internal data shape |
| C. Out-of-scope observations | New concerns / risks / opportunities surfaced during implementation but not in PLAN/SPEC | "Should we also rate-limit /api/profile?", performance follow-up |
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.
- 2d ago First seen · 178 lines · 62 tokens per session scan A 1817a2a22803
reviewer is an agent published in the GitHub repository howar31/magi-workflow (2 stars, last pushed 3d ago), licensed MIT. It adds 62 tokens to every session and 1,691 once invoked, about $0.0003 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
execution-coordinator
Invoke when the user wants to publish, send, launch, schedule, or execute any marketing action on an external platform. Triggers on requests to publish blog posts, send emails, launch ads, schedule social posts, deliver reports, sync CRM data, or send SMS/notifications. Manages the approval workflow and ensures every…
content-producer
Agent san xuat noi dung — viet script, copy, brief creator, lap lich noi dung.
accessibility-specialist
Accessibility expert: WCAG 2.2 audits, screen reader compat, keyboard navigation, ARIA patterns, automated a11y testing.
corporate-strategy
Corporate Strategy (CSO). Owns plugins/corporate-strategy/ and nothing else. Delegate work in this department's remit here.
demand-generation
Demand Generation (CMO). Owns plugins/demand-generation/ and nothing else. Delegate work in this department's remit here.
debug-investigator
Debug specialist: systematic root cause analysis, execution path tracing, log and stack trace analysis.