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 davidYichengWei/Agency --skill process-cr-commentsgit clone --depth 1 https://github.com/davidYichengWei/AgencyWrote 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/davidyichengwei/agency/process-cr-comments)<a href="https://agentmods.dev/skills/davidyichengwei/agency/process-cr-comments"><img src="https://agentmods.dev/badge/skills/davidyichengwei/agency/process-cr-comments.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 | $0.00212 | $0.03932 |
| Opus 5 | $0.00106 | $0.01966 |
| Sonnet 5 | $0.00042 | $0.00786 |
| Haiku 4.5 | $0.00021 | $0.00393 |
Grade A, and why
process-cr-comments 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 6d 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 — 216 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Process Review Comments
Review comments — whether from a human reviewer, an AI reviewer in CI, or a design-doc commenter — often mix real issues with things that won't survive a fresh read of the code or spec. Acting on everything bloats the diff; dismissing silently loses the reviewer's trust. This skill does the middle thing: validates each finding independently, fixes what holds up, and produces a reply that gives the reviewer enough context to approve.
The report's purpose is consensus. It's not a private log — it's the message you'll send back to the reviewer so they can click "Approve". That shapes two things:
- For every fix, explain how you fixed it — enough detail that the reviewer can confirm the fix matches what they were worried about, without re-reading the whole diff.
- For every non-fix (dismissed, downgraded, deferred, stylistic-only, not-this-MR), explain why in terms the reviewer can accept — specific counter-evidence, a citation, a design rationale, or a scope boundary. A reviewer won't approve on "we decided not to" alone; they will approve on "here's why that concern doesn't apply here."
A finding is only real if a fresh read of the code or spec can reproduce it. The skill does exactly that — in parallel, one validator per finding — before deciding anything.
Workflow
- Ingest review comments → parse into discrete findings
- Gather scope → changed files / changed spec sections, PR or doc intent, related spec
- Dispatch validators in parallel → one
validatorsubagent per finding - Final judgement → main agent reconciles verdicts, handles disagreements, dedups
- Fix confirmed issues → apply the suggestions (or write fresh fixes) in this session
- Reply report → consensus-oriented markdown, one entry per finding, written to the reviewer
Step 1 — Ingest review comments
Review comments arrive in a few shapes. Detect which and normalize into a list of findings.
Shape A — An MR / PR / ticket URL or ID. The user points at a hosted review (e.g. an MR URL, a PR number, a review ticket ID). Use whatever MCP or CLI the environment provides to fetch (a) the comments on the review — human and AI-authored — and (b) the diff or changed files, so validators can read current code. Don't assume a specific tool; pick the one that matches the host.
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.
- 6d ago First seen · 216 lines · 212 tokens per session scan A 09ef61bd4b34
process-cr-comments is a skill published in the GitHub repository davidYichengWei/Agency (5 stars, last pushed 4mo ago), licensed MIT. It adds 212 tokens to every session and 3,932 once invoked, about $0.0011 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
autoreview
Pre-commit/ship code review: Codex default; optional Claude or Pi.
rework-rate
Measure and interpret PR rework rate — the emerging 5th DORA metric.
omh-code-review
This is a Hermes-native code-review workflow skill.
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"…
code-reviewer
Code review specialist focused on patterns, bugs, security, and performance.
full-repo-review
Comprehensive four-wave review of all repo source files, producing a prioritized issue backlog.