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 skills/guillermoscript/lms-front/pr-review-loopnpx skills add guillermoscript/lms-front --skill pr-review-loopgit clone --depth 1 https://github.com/guillermoscript/lms-frontWhat 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 | $0.00188 | $0.03975 |
| Opus 5 | $0.00094 | $0.01988 |
| Sonnet 5 | $0.00038 | $0.00795 |
| Haiku 4.5 | $0.00019 | $0.00398 |
Grade C, and why
pr-review-loop scanned grade C with 1 finding 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.
Recursive force deletehighDestructive command
rm -rf with a variable or a broad path is one typo away from removing the wrong tree.
`rm -rf`. If cleanup skipped something, the merge still succeeded — say How it starts
The opening of the file, as written. The whole thing — 334 lines — stays where its author put it; the contents beside it link to each section on GitHub.
PR Review Loop — from announced PR to merged, closed out
The gap this closes: after ship-pr announces a PR, nothing watches it.
This skill polls the PR for reviewer feedback, addresses each piece with
code or an answer, loops until the PR is approved, merges it, and triggers
ship-pr's close-out — the whole review conversation handled without the
user shepherding it.
Two doctrines govern everything below:
- Stateless — GitHub is the only state. No watermark files, no local memory of what was processed. Each cycle rebuilds its work queue from the PR itself: a review thread is pending if it's unresolved and the last word isn't ours; a conversation comment is pending if it's newer than our last reply or push and isn't ours. A dead session, an expired cron, or a manual re-invoke all resume identically — just run a cycle.
- The loop never reacts to its own comments. "Ours" is the
authenticated user (
gh api user -q .login), resolved fresh each run.
Config (Slack channel, default reviewer, project board) comes from
.claude/gh-workflow.config.json via the gh-repo-config skill; board
moves go through gh-board; close-out comments belong to
ship-pr; local cleanup after the merge runs through this skill's own
scripts/cleanup.sh. Chat with the user may be
terse (caveman), but everything posted to GitHub or Slack is a permanent
record — normal, professional, full-sentence English.
Reviewing someone else's PR is the mirror skill, pr-review-watch —
it runs /code-review, submits an approve/request-changes verdict, and
re-reviews each new push. This skill is only for PRs we authored; one PR
never gets both loops.
Invocation and arming
/pr-review-loop <pr-url | #N | N>
/pr-review-loop # resolves the PR from the current branch
With no argument, resolve via gh pr view --json number,url on the
current branch. Verify the PR is open and authored by (or assigned to)
the authenticated user — this skill shepherds our PRs, not strangers'.
What ships with it
2 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
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.
- 3d ago First seen · 334 lines · 188 tokens per session scan C 90bcd24c9b07
pr-review-loop is a skill published in the GitHub repository guillermoscript/lms-front (24 stars, last pushed 3d ago), licensed MIT. It adds 188 tokens to every session and 3,975 once invoked, about $0.0009 per session on Opus 5. A static security scan graded it C with 1 finding (recursive force delete). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other skills, from other repositories
api-development
FastGPT API 开发规范。重点强调使用 zod schema 定义入参和出参,在 API 文档中声明路由信息,编写对应的 OpenAPI 文档,以及在 API 路由中使用 schema.parse 进行验证。.
ci-workflow-sync
FastGPT CI workflow 双轨同步。当用户修改或新增 .github/workflows/ 下的 GitHub Actions workflow 时必须触发：同步更新 .forgejo/workflows/ 对应文件保持功能一致，或判断是否需要新建 Forgejo 版本。涉及 CI、GitHub Actions、Forgejo Actions、镜像构建、container registry、artifact、workflow yaml 改动、build- workflow、test- workflow 时也使用此技能。即使用户只提到"改一下 CI"或"加个 workflow"也应触发。.
prompt-optimize
Expert prompt engineering skill that transforms Claude into "Alpha-Prompt" - a master prompt engineer who collaboratively crafts high-quality prompts through flexible dialogue. Activates when user asks to "optimize prompt", "improve system instruction", "enhance AI instruction", or mentions prompt engineering tasks.
deprecate-workflow-node
当用户需要弃用一个工作流节点（保留向后兼容、隐藏出模板面板）时触发该 skill。FastGPT 工作流节点的弃用流程标准化封装，覆盖模板、Dispatcher、UI 引用等所有需要改动的位置。.
doc-i18n
将 FastGPT 文档从中文翻译为面向北美用户的英文。当用户提到翻译文档、i18n、国际化、translate docs、新增/修改了中文文档需要同步英文版时，使用此 skill。也适用于用户要求检查文档翻译缺失、批量翻译、或对比中英文文档差异的场景。.
pr-change-analysis
手动触发的 FastGPT PR 或本地分支变更梳理技能。仅当用户显式调用 $pr-change-analysis 时使用；用于 reviewer 分析一个 GitHub PR 或当前本地分支相对 upstream/main 的需求变更、影响范围、代码质量与代码风格，不用于自动审查触发。.