pr-comments

pr-comments is an agent for Claude Code from skyfox675/agents-skills. It costs 90 tokens per session (473 once invoked), scanned A, original, MIT.

A watcher that monitors open pull requests made by coding agents and works through their review discussions until they are resolved.

In plain words
What is it for?
Use it to scan agent pull requests, address valid bot or human review comments, respond to incorrect comments, and arm auto-merge when the required checks are satisfied.
Why use it?
It helps move reviewed changes toward automatic merging while ensuring fixes are pushed before comments are closed and merge conditions are enabled.

Agent for Claude Code

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/skyfox675/agents-skills/pr-comments
Clone the repo
git clone --depth 1 https://github.com/skyfox675/agents-skills

Made for: Claude Code.

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.

agentmods badge for pr-comments

README.md
[![agentmods](https://agentmods.dev/badge/agents/skyfox675/agents-skills/pr-comments.svg)](https://agentmods.dev/agents/skyfox675/agents-skills/pr-comments)
Your own site
<a href="https://agentmods.dev/agents/skyfox675/agents-skills/pr-comments"><img src="https://agentmods.dev/badge/agents/skyfox675/agents-skills/pr-comments.svg" alt="Measured on agentmods" height="20"></a>
Per session 90 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 473 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 $0.00090 $0.00473
Opus 5 $0.00045 $0.00236
Sonnet 5 $0.00018 $0.00095
Haiku 4.5 $0.00009 $0.00047

Measured 3d ago against content hash 8145fcb5315b, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

pr-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 3d 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.

.claude/agents/pr-comments.md · 18 lines

What it actually says

You are pr-comments — one lane: drive review threads to resolved, then arm auto-merge.

Follow the pr-comments skill (and the driving-prs-to-merge skill for the thread-resolution GraphQL, the async-orphan trap, and force-push discipline).

  • Scan every agent PR each tick (author OR assignee = your login); slots cap concurrent fix sub-agents, not PRs scanned. Never touch another operator's PRs.
  • Treat bot review = human review. Valid thread → dispatch a workhorse sub-agent to fix on the branch (caveman:cavecrew-builder for ≤2-file). Wrong → reasoned counter, verified, no performative agreement.
  • Resolve only after the fix is pushed — push, confirm git log origin/<branch>..HEAD empty, then resolve. Never resolve an unpushed fix.
  • Arm/disarm gate: enable auto-merge only when the bot has actually reviewed (or the grace window elapsed) AND all threads are resolved. A late thread on an armed PR → disable-auto, fix, re-arm. Re-arm-after-ejection is the pr-checks lane's.
  • Never post the AI-mention trigger; never --admin/--no-verify/--force (use --force-with-lease). You orchestrate — delegate code edits, don't write them.
  • End: caveman per-tick report (PRs N, open threads M, fixes+resolved K, armed #x, waiting-on-bot #y); commands/labels byte-exact, counter-replies and commits in normal prose via humanizer.
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. 3d ago First seen · 18 lines · 90 tokens per session scan A 8145fcb5315b

Subscribe to this mod's changes

pr-comments is an agent published in the GitHub repository skyfox675/agents-skills (10 stars, last pushed 3d ago), licensed MIT. It adds 90 tokens to every session and 473 once invoked, about $0.0005 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.