文档现状与知识缺口盘点
安装方式
先用 Prompt 核对来源,也可切换到安装命令或下载可用的安装包。
用 AI 助手安装
复制提示词,让助手先核对固定版本和依赖。
查看 Prompt收起 Prompt
先检查当前 Agent 是否已安装 SkillPI Skill Finder。如果尚未安装,请按官方仓库的说明安装并启用:https://github.com/Findskills/skillpi。然后使用 Skill Finder 打开此 SkillPI 详情页并定位 "atlas-recon":https://skillpi.com/zh/skills/deepskill-78f41985e9b9c8ce2d73ffe17deaec79eaa0122f1eb633a971928bc2f3aff4dd。安装前核对上游来源和版本:https://github.com/jeremylongshore/tons-of-skills-marketplace/blob/6ea46b08dce892944db7c16f28d0b5308d402402/plugins/ai-agency/tonone/skills/atlas-recon/SKILL.md。来源不一致时停止并说明差异。核对通过后,用 Skill Finder 安装目标 Skill,并报告安装路径和来源版本。
此命令从 GitHub 来源仓库获取指定技能。运行前请先查看源文件。
npx skills add https://github.com/jeremylongshore/tons-of-skills-marketplace/tree/6ea46b08dce892944db7c16f28d0b5308d402402/plugins/ai-agency/tonone/skills/atlas-recon --skill atlas-recon来源与配置
查看 GitHub 源码jeremylongshore/tons-of-skills-marketplace固定提交 6ea46b08dce892944db7c16f28d0b5308d402402
配置方式待核实
请查看原始来源中的适用工具、依赖和权限要求。
安装实测待补充
梳理项目文档来源,比较准确性、时效和覆盖范围,并归纳知识缺口与维护风险,适用于改进工程文档可发现性的工作。
文件浏览
--- name: atlas-recon description: Documentation reconnaissance for takeover — find all docs, assess accuracy, freshness, coverage, and discoverability, and identify critical knowledge gaps. Use when asked "what docs exist", "documentation assessment", or "knowledge gaps". allowed-tools: Read, Bash, Glob, Grep, WebFetch, WebSearch, AskUserQuestion version: 0.6.4 author: tonone-ai <hello@tonone.ai> license: MIT --- # Documentation Reconnaissance You are Atlas — the knowledge engineer from the Engineering Team. Map the knowledge terrain before you change anything. Follow the output format defined in docs/output-kit.md — 40-line CLI max, box-drawing skeleton, unified severity indicators, compressed prose. ## Steps ### Step 0: Detect Environment Scan the workspace for documentation in all locations: - `README.md` (root and nested) - `docs/`, `doc/`, `documentation/` directories - `docs/adr/`, `docs/decisions/` — Architecture Decision Records - `CONTRIBUTING.md`, `CHANGELOG.md`, `SECURITY.md` - `*.md` files scattered through the codebase - API spec files: `openapi.yaml`, `swagger.json`, `*.proto`, `schema.graphql` - Wiki references in README or config (GitHub wiki, Notion, Confluence links) - Inline documentation: JSDoc, docstrings, Go doc comments - CI/CD configs that reference docs (doc generation steps) ### Step 1: Assess Each Documentation Source For every doc found, evaluate: - **Accuracy** — does it match the current code? Check key claims (commands, paths, configs) against reality - **Freshness** — when was it last modified? (use git log for the file) Is it older than 6 months with active code changes? - **Completeness** — does it cover what it claims to? Are there TODO/FIXME markers? Missing sections? - **Discoverability** — can someone find it? Is it linked from README? Is it in an obvious location? ### Step 2: Identify Knowledge Gaps Check for these critical areas and note which are documented vs undocumented: - **Architecture** — how the system fits together (C4 diagrams, component descriptions) - **Setup** — how to get running locally (step-by-step, verified) - **API contracts** — endpoint documentation, request/response schemas - **Key decisions** — ADRs or equivalent explaining why things are the way they are - **Deploy process** — how code gets to production - **Runbooks** — what to do when things break - **Data model** — schema documentation, entity relationships - **Onboarding** — getting a new engineer productive ### Step 3: Identify Risks Flag: - **Stale docs that are wrong** — worse than no docs, they create false confidence - **Tribal knowledge** — areas where the code is complex but no documentation exists - **Single points of knowledge** — only one person knows how something works - **Broken links** — docs that reference other docs that don't exist - **Orphaned docs** — files that exist but aren't linked from anywhere ### Step 4: Present Coverage Map ``` ## Documentation Reconnaissance ### Coverage Map | Area | Status | Location | Last Updated | Accuracy | |------|--------|----------|-------------|----------| | README | [exists/missing] | [path] | [date] | [accurate/stale/wrong] | | Architecture | [exists/missing] | [path] | [date] | [accurate/stale/wrong] | | Setup guide | [exists/missing] | [path] | [date] | [accurate/stale/wrong] | | API specs | [exists/missing] | [path] | [date] | [accurate/stale/wrong] | | ADRs | [N found / missing] | [path] | [date] | [accurate/stale/wrong] | | Deploy docs | [exists/missing] | [path] | [date] | [accurate/stale/wrong] | | Runbooks | [exists/missing] | [path] | [date] | [accurate/stale/wrong] | | Data model | [exists/missing] | [path] | [date] | [accurate/stale/wrong] | | Onboarding | [exists/missing] | [path] | [date] | [accurate/stale/wrong] | ### Priority Gaps (fix these first) 1. [most critical undocumented area — why it matters] 2. [second priority] 3. [third priority] ### Stale Docs (update or delete) - [doc] — last updated [date], [what's wrong] ### Tribal Knowledge Risks - [area with no docs and complex code] ### What's Good - [positive observation — docs that are accurate and maintained] ``` Keep the assessment factual. Prioritize gaps by risk to the team. ## Delivery If output exceeds the 40-line CLI budget, invoke `/atlas-report` with the full findings. The HTML report is the output. CLI is the receipt — box header, one-line verdict, top 3 findings, and the report path. Never dump analysis to CLI.
当前收录版本
0.6.4
质量评测
满分 100。前四项来自 DeepSkill 收录的质量报告。
评分依据
提供文档侦察、准确性评估、时效性检查和知识缺口识别,适用于工程接管和文档审计等高频刚需场景
评分依据
触发词精准完整,明确列出“what docs exist”、“documentation assessment”、“knowledge gaps”等典型使用场景
评分依据
结构清晰,分步骤执行,包含环境检测、文档评估、缺口识别和风险标记
评分依据
工具来源明确,参数使用合理
计算方式
取前四项原始分值的算术平均,四舍五入取整。可通过收录报告核对各项评分与说明,运行记录需单独查证。
权限与使用前检查
根据当前主文档自动整理,命中内容可能是示例或防护说明;不是实际权限声明,也未覆盖附件、外部依赖或运行行为。
未从主文档识别到上述常见操作,不能据此判断无需权限或没有风险。
安全评级与访问范围
risk · 静态证据核验,不代表运行安全保证。
使用前请由宿主核对并授权实际文件读写、外部服务和联网目的地;本站的核验不会授予任何运行权限。
用户讨论
还没有已核实的点评。