Token导航 LogoToken导航

Agent Skill · 研究检索

repo-sentinel(回购哨兵)

repo-sentinel 用于查找、检索和筛选相关信息,适合在 Codex、Claude、Cursor、Gemini CLI 中需要根据关键词、任务场景或来源线索快速定位候选结果时使用。可结合来源仓库、安装命令和原始 README 继续核验具体用法。安装前建议确认权限范围、维护状态,以及是否会触发联网、命令执行或文件读写。

repo-sentinel 是研究检索类 Agent Skill,可用宿主为 Codex、Claude、Cursor、Gemini CLI,通过GitHub 仓库获取,权限为接触敏感数据;许可证信息还不明确,第三方扫描有未通过项,按安装量在研究检索类的 73,675 个 Skill 里排第 44,490。

总安装 912周安装 38GitHub Stars 216下载 304核验 2026-05-01

安装说明

本站只整理中文说明,不托管安装包,也不代用户安装。
GitHub
许可证
unknown

安装方式

通过对话安装

复制提示词发给支持本地命令或 Skills 的 AI 助手,先确认命令和权限,再让它执行。

请帮我安装这个 Agent Skill:repo-sentinel(回购哨兵)
来源仓库:https://github.com/mathews-tom/armory
仓库路径:skills/repo-sentinel
安装命令:
npx skills add https://github.com/mathews-tom/armory --skill repo-sentinel
安装前请先检查当前环境是否支持对应 CLI,并向我确认将要执行的命令、安装目录、联网范围和文件读写权限;确认后再执行。

命令行安装

复制命令到本机终端执行。该命令会通过 npx skills 从第三方来源获取 Skill;本站只展示命令,不托管安装包,也不自动执行。

skills.shnpx skills
npx skills add https://github.com/mathews-tom/armory --skill repo-sentinel

这个 Skill 是做什么的

研究检索类 Skill 主要解决「找资料、核对事实、把来源整理清楚」的问题,通常会让助手先检索再总结。

repo-sentinel 用于查找、检索和筛选相关信息。适合根据关键词、任务场景或来源线索快速定位候选结果。适用于 Codex、Claude、Cursor、Gemini CLI 中的研究类任务。通过 npx 安装,需确认权限范围和维护状态,注意可能触发联网操作。

建议结合来源仓库和原始 README 核验具体用法,避免依赖未经验证的搜索结果。

来源仓库内路径:skills/repo-sentinel

上手步骤

4 步
  1. 1

    确认自己的宿主工具

    公开资料标注支持 Codex、Claude、Cursor、Gemini CLI;先确认自己用的是其中之一,再按对应工具的 Skills 目录放置。

  2. 2

    拿到安装命令

    可以直接复制的命令是 npx skills add https://github.com/mathews-tom/armory --skill repo-sentinel;命令里的占位符换成自己的路径或密钥再执行。

  3. 3

    安装后重启宿主

    命令执行完要完全退出宿主工具再打开,让新装的 Skill 被重新扫描到,否则列表里可能看不到。

  4. 4

    按最小权限试用

    它需要接触敏感数据,先在测试目录或测试账号里跑,确认没问题再放到正式项目。

常见问题

6 条

repo-sentinel 怎么安装?

在终端执行 npx skills add https://github.com/mathews-tom/armory --skill repo-sentinel。装完完全退出宿主工具再打开,让它重新扫描本地 Skill。

repo-sentinel 支持哪些 AI 工具?

公开资料标注支持 Codex、Claude、Cursor、Gemini CLI。其他工具只要能读取本地 Skill 目录,一般也能手动放进去,但要自己确认触发方式是否一致。

repo-sentinel 需要什么权限?

权限级别是接触敏感数据。该 Skill 可能接触密钥、Token、环境变量或敏感配置,应进入高风险复核队列,默认不自动发布。

repo-sentinel 是免费的吗?

公开资料没有标注明确许可证,收费和商用限制都无法确认。正式使用前请到来源仓库核对授权说明。

repo-sentinel 用的人多吗?

公开数据是累计安装 912 次,近一周安装 38 次,下载 304 次,GitHub 216 星。这些数字来自第三方目录统计,只能当作热度参考,不代表质量或安全性。

repo-sentinel 安全吗?

第三方扫描有未通过项,风险较高。除非你确认过源码,否则不建议直接安装到常用环境。

值得选的地方和要先想清楚的地方

值得选的地方

  • 同时适配 4 种工具,换客户端不用重新找替代品。

先想清楚的地方

  • 需要接触敏感数据,装到常用环境前要先评估影响面。
  • 许可证没有明确标注,能不能商用、能不能再分发都不确定。
  • 第三方扫描存在提醒项或未通过项,需要人工逐条确认。
  • 第三方目录整理,版本更新和行为细节以来源仓库为准。

适不适合你

更适合

  • 已经在用 Codex,想直接扩展它能力的人。
  • 需要处理研究检索类任务、又不想每次都手写提示词的人。

不太适合

  • 环境里有大量密钥、凭证,不方便给第三方 Skill 读取的人。
  • 需要明确授权才能用到公司项目里的场景。
  • 需要厂商级技术支持和 SLA 的生产系统。

研究检索类 Skill 对比

6 个同类
Skill总安装GitHub 星权限许可证扫描
find-skills

寻找技能 · Codex、Claude

31,199,81316,486操作浏览器MIT有提醒
self-improving-agent

自我改进 Agent · OpenClaw

16,464,0023,396接触敏感数据MIT-0通过
soultrace

灵魂追踪 · Codex、Claude

7,588,94412需要联网未明确有提醒
self-improving

自我完善 · OpenClaw

6,829,2841,046需要联网MIT-0通过
microsoft-foundry

微软代工厂 · Codex、Claude

6,614,144764调用外部服务未明确有提醒
repo-sentinel

回购哨兵 · 本页

912216接触敏感数据未明确未通过

按安装量取同分类前几名;数字来自第三方目录统计,只作热度参考,不代表质量结论。

SKILL.md 原文

Repo Sentinel

Everything in a public repo is permanent attacker surface. This skill defines what belongs in a public repo, what does not, how to detect violations across 12 attack surfaces, how to remediate when the boundary is violated, and how to enforce continuously.

Reference files

This skill uses bundled reference files for detailed patterns and templates. Read them as needed:

FileWhen to read
references/attack-surfaces.mdWhen auditing any surface — full definitions for Surfaces 0–12
references/scan-patterns.mdWhen running any audit (fast-path or full) — contains all detection commands
references/pre-release-checklist.mdWhen running the Pre-Release Audit (Stage 4) — §4.1–§4.8 readiness checklist
references/templates.mdWhen setting up enforcement, generating.gitignore, or creating CI gates
references/remediation.mdWhen fixing findings or scrubbing history — contains all fix procedures

Prerequisites

  • gh CLI installed and authenticated (gh auth status must pass) — required for GitHub-specific surface checks (Surface 10)
  • Active git repository context — the skill operates on git objects; non-git directories are out of scope
  • trufflehog or gitleaks — optional but strongly recommended for Surface 0 (git history) secret detection with entropy analysis; without them, fall back to git log -p grep patterns from references/scan-patterns.md
  • Read access to the full git object store — shallow clones (--depth N) will miss history secrets; warn the user if a shallow clone is detected

Calibration Rules

  • Public vs. private visibility: Apply stricter severity ratings for public repos — findings classified MEDIUM in a private repo (e.g., internal URL in a comment) escalate to HIGH in a public repo. Confirm repo visibility before scoring.
  • Stack-scoped surfaces: Scope the audit to attack surfaces relevant to the detected tech stack. A static HTML repo has no meaningful Surface 6 (containers) or Surface 7 (lock files) exposure — mark those surfaces N/A rather than penalizing.
  • N/A handling: Surfaces scored N/A are not penalized and do not lower the overall risk posture. Document N/A surfaces explicitly so the user understands what was skipped.
  • Tool availability: If trufflehog/gitleaks are unavailable, note this in the audit header and describe the reduced confidence in Surface 0 coverage.
  • False positive discipline: Flag a finding only when there is evidence of actual exposure, not just pattern proximity. A variable named api_key with a placeholder value is LOW, not CRITICAL.

Foundational Principle

The public/private boundary is a one-way valve. Once a byte reaches a public remote — via push, PR, issue, wiki, release asset, or GitHub Pages — assume it is indexed, cached, mirrored, and archived permanently. git push --force, PR deletion, issue edits, and release removal do NOT guarantee erasure. Scraping infrastructure (GitHub Archive, GH Torrent, Software Heritage, Google Cache, Wayback Machine, and dozens of proprietary security scanners) operates continuously with sub-hour latency.

Decision framework for every artifact:

QuestionIf YES →If NO →
Could this help an attacker who has no other access?EXCLUDEContinue
Does this reveal internal topology not inferable from public signals?EXCLUDEContinue
Does this contain values that grant access to anything?EXCLUDEContinue
Does this violate a license obligation or expose legal risk?EXCLUDEContinue
Would removing this reduce the repo's utility to legitimate users?INCLUDE (if above = all NO)EXCLUDE

When in doubt, exclude. False negatives (leaked secrets) are catastrophic and irreversible. False positives (over-redaction) are trivially correctable.


The 12 Attack Surfaces

Full definitions — what belongs, what doesn't, why it leaks, detection approach — for Surfaces 0 through 12 are in references/attack-surfaces.md. Read that file when scoping or scoring any audit. Scan commands per surface live in references/scan-patterns.md; remediation procedures in references/remediation.md.


Severity Classification

All findings are classified by severity. The classification drives action priority:

SeverityCriteriaAction
CRITICALActive credential exposure, private key, auth tokenBlock push. Fix immediately.
HIGHInfrastructure/topology enabling targeted attackResolve before push.
MEDIUMInformation leakage aiding reconnaissanceFix in next commit.
LOWHygiene, style, redundancy issuesFix at convenience.

CRITICAL and HIGH in git history → full history scrub + credential rotation required.


Operations

Fast-Path Audit (Staged Changes Only)

Use when pushing a single file or small changeset. Scans only staged changes, not the full repo. Read references/scan-patterns.md § Fast-Path for the commands.

Full Repo Audit (20+ checks)

Run before making any repo public or before first push to a public remote. Read references/scan-patterns.md § Full Audit for the complete 20-check sequence, the Quick-Reference Scan Commands block, and the audit output format.

Pre-Release Audit Mode (4-Stage DAG)

When preparing a repo for open-source release, run this 4-stage pre-release audit instead of the surface-based audit. Each stage emits PASS / WARN / FAIL with actionable remediation. Hard blockers in stages 1–3 halt the pipeline. Stage 4 produces advisory output.

Stage 1: Sensitive Assets        [HARD BLOCKER] → Surfaces 0–4, 8–9
Stage 2: Legal & Compliance      [HARD BLOCKER] → Surface 11
Stage 3: Public Surface Hygiene  [HARD BLOCKER] → Surfaces 4–7, 9–10
Stage 4: Contribution & Release  [SOFT BLOCKER] → Surface 12 + Pre-Release Checklist

Run stages sequentially. Report results in a structured audit table at the end. Stage 4 checklist items (§4.1–§4.8) are in references/pre-release-checklist.md.

Continuous Enforcement Setup

Shift-left prevention is the highest-leverage action. Read references/templates.md for ready-to-use pre-commit config, GitHub Actions workflow, and.gitignore generator.

History Contamination Remediation

When secrets have already been committed. Read references/remediation.md for the full triage decision tree, git filter-repo commands, BFG fallback, post-scrub protocol, and.gitignore generation guidance.


Limitations

  • History scrubbing does not guarantee removal of exposure. Force-push is required, and external mirrors (forks, GitHub Archive, Software Heritage) retain history indefinitely regardless of local operations.
  • External mirrors, caches, and search engine indexes cannot be verified as de-indexed after content removal.
  • Single-repo scope only — not designed for monorepo audits without adaptation. Cross-package secret propagation requires separate analysis per package root.
  • GitHub-specific checks (branch protection, secret scanning alerts, security advisories) require the gh CLI with authenticated access. Without it, Surface 10 coverage is reduced.
  • Secret scanning depth depends on available tooling. trufflehog and gitleaks provide verified detection with entropy analysis; manual regex patterns used as fallback have higher false-positive rates and miss obfuscated credentials.
  • Artifact decisions for package registry publishing (npm, PyPI, crates) have ecosystem-specific norms that differ from source repo inclusion rules — apply ecosystem conventions when auditing published artifacts.

平台分布

Codex

37.93%
按下载量换算115

Claude

28.94%
按下载量换算88

Cursor

17.27%
按下载量换算53

Gemini CLI

9.54%
按下载量换算29

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

未通过

权限和风险

接触敏感数据

该 Skill 可能接触密钥、Token、环境变量或敏感配置,应进入高风险复核队列,默认不自动发布。

安装前确认

本站仅展示第三方公开信息,不托管安装包,不提供自动安装或运行环境。安装前应自行审查源码、依赖和命令行为。来源安全扫描存在提醒或未通过项,安装前请自行复核。目前只有一个公开来源,建议同时核对源仓库。

继续浏览同类 Skills