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 MadAppGang/magus --skill aggregate-reviewsgit clone --depth 1 https://github.com/MadAppGang/magusWrote 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/madappgang/magus/aggregate-reviews)<a href="https://agentmods.dev/skills/madappgang/magus/aggregate-reviews"><img src="https://agentmods.dev/badge/skills/madappgang/magus/aggregate-reviews/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/madappgang/magus/aggregate-reviews"><img src="https://agentmods.dev/badge/skills/madappgang/magus/aggregate-reviews.svg" alt="Reviewed on agentmods" width="80" 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 | $0.00047 | $0.05171 |
| Opus 5.5 | $0.00019 | $0.02068 |
| Sonnet 5.5 | $0.00009 | $0.01034 |
| Haiku 4.5 | $0.00005 | $0.00517 |
Grade A, and why
aggregate-reviews 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 22d 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 — 389 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Aggregating reviews and research findings
Turn several independent reports into one, marking where they agree and where they diverge, and compute the verdict the dispatcher's thresholds define. You are handed reports, never the thing they judged. You do not review, you do not re-derive findings from code, and you do not soften or sharpen what a reviewer said. You count it.
Two forms
Decide the form from the first lines of the prompt.
| Form | Marker | Dispatched by | Follow |
|---|---|---|---|
| Review consolidation | a REVIEWS: line |
/dev:audit, /dev:fix Phase B, Phase 3 plan review, Phase 5 code review, the multimodel skills |
Review consolidation |
| Research synthesis | SESSION_PATH: with ITERATION: or MODE: final_report, and no REVIEWS: |
/dev:research |
Research synthesis |
A SESSION_PATH: line beside REVIEWS: is context only; REVIEWS: wins.
Two rules that hold in both forms
Given reports, never code. Read only the paths the prompt names. Do not open the files a review talks about, do not run a capture script, do not grep the codebase to check whether a reviewer was right. The moment you look at code you become a reviewer whose findings nobody else saw, and the consolidated report stops being a consolidation. A finding you doubt is reported with its consensus level. It is not dropped, and not re-investigated.
No rules of your own. The severity scale and the verdict thresholds belong to the
reviewer that produced the reports. The dispatcher quotes the thresholds on
THRESHOLDS:; apply those words and nothing else. Never recall a threshold from memory,
never invent one when the line is missing, never restate one as if it were yours. Cite
it as "per THRESHOLDS". If THRESHOLDS says WARN, the verdict is WARN.
Review consolidation
Contract
REVIEWS: one path per line (absolute, or relative to the working directory).
The internal review first, then one per external slot that completed.
Every line is an input you must account for.
THRESHOLDS: the verdict rules, quoted verbatim by the dispatcher: each rule a verdict
word and the condition that earns it, in the vocabulary of the reviewer
whose reports these are. A code review quotes dev:reviewer's three
lines; a plan review states Phase 3's own rule; a design review passes
the designer's percentage scale; a docs audit passes dev:docs' score
bands; a /dev:fix patch vote passes fix.md's report-to-vote mapping and
consensus rules.
OUTPUT: path of the consolidated report. You write exactly this one file.
N: optional; the panel size a vote tally is over. Only a /dev:fix patch
vote carries it: every slot is a line under REVIEWS whether or not its
file exists, so N equals the line count.
SESSION_PATH: optional context. Not a file to read.
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.
- 22d ago Changed · +5 lines a09846cd7c14
- 1mo ago First seen · 384 lines · 47 tokens per session scan A ac872672ec95
aggregate-reviews is a skill published in the GitHub repository MadAppGang/magus (13 stars, last pushed today), licensed MIT. It adds 47 tokens to every session and 5,171 once invoked, about $0.0002 per session on Opus 5.5. 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 2026-09-15.
Other skills, from other repositories
autoreview
Structured code review through the shared OpenClaw agent-skills installation.
omh-code-review
This is an OMH code-review workflow skill, projected for Agent Skills hosts (Claude Code, Codex, Cursor, opencode, OpenClaw, pi).
revdiff-plan
Review the last Codex assistant message (plan, analysis, or proposal) with inline annotations in a TUI overlay. Extracts the most recent response from Codex rollout files and opens it in revdiff for review and annotation. Activates on "revdiff-plan", "review plan with revdiff", "annotate plan", "review last response"…
asplos-writing-style
Use when drafting or revising ASPLOS prose — engineering the first two pages that rapid review actually reads, stating the cross-layer insight as one quotable sentence, structuring per-layer mechanism sections, writing bounded claims, and fitting everything into 11 self-contained pages that include figures and…
explore-unknowns
Guide the user through a quadrant walk that maps the unknowns of a task — open by listing the known knowns, then work through known unknowns, unknown knowns, and unknown unknowns one stage at a time, ending with a complete four-quadrant map in the user's hands. Use when a request is ambiguous or underspecified, the…
re0-merge
Review and land an external contribution the way this suite does: gate it against the thesis, land it with the author's credit intact, complete a new skill rather than merging it raw, then approve, credit, and explain before closing. Use when reviewing a pull request, as any collaborator or maintainer, not only the…