SuperAPI AI Directory

baiducloud-docs

General & other

View repository

Updated 11-10 06:15

About this skill

BaiduCloud Docs

Use this Skill as the shared official documentation source for every Baidu Cloud task. Other Skills orchestrate operations; when they need a product explanation, parameter meaning, limit, or error explanation, retrieve it here instead of relying on memory or text copied into a Skill.

All commands are read-only HTTPS requests to the public documentation site (cloud.baidu.com and bce.bdstatic.com). They need no login, profile, or credentials, and output JSON.

Choose the command

  • search <query...> — natural-language search across all products. Use it when the target document is unknown. Add --product <code> to keep only one product's documents; including the product name in the query also helps.
  • get <doc_id> — read one document's full Markdown body plus title, update date, and URL. Long documents are cut at a section boundary; continue with --offset <next_offset> or jump to an offset listed in headings.
  • tree --product <code> [<doc_id>] — browse a product's table of contents. --depth (default 1) limits expansion; pass a folder node's doc_id to expand only that branch instead of the whole tree.
  • grep --product <code> <keyword...> — keyword hits inside one product, for configuration items, error codes, and terms. Pass separate words as separate arguments (grep --product BOS Bucket Policy); each must appear. Quote an argument to require it as one contiguous phrase, for example an exact error message (grep --product BCC "does not exist"). Hits whose snippet contains every keyword come first with exact: true; remaining slots are filled with exact: false candidates, whose snippet lacks some keyword.
  • products [term] — list product codes (for example 函数 → CFC). Product codes are case-sensitive in tree and grep.

doc_id is <product>/<id> (for example CFC/gkwfyd83w). get also accepts the document URL https://cloud.baidu.com/doc/<product>/s/<id>.

node scripts/docs.mjs search 函数计算 冷启动优化 --limit 3
node scripts/docs.mjs get CFC/gkwfyd83w
node scripts/docs.mjs tree --product CFC --depth 2
node scripts/docs.mjs grep --product CFC 超时时间
node scripts/docs.mjs products 对象存储

Typical flow: search (or products + tree) to locate a document, then get to read it. Use grep when the user already named the product and an exact term or error code.

Answering rules

  • Base the answer on the retrieved text and cite each source with its url. Mention the document updated date when timeliness matters.
  • Read the document with get before quoting parameters, limits, or steps; search summaries are excerpts, not the full context. Treat a grep hit with exact: false as a lead only: cite it after get confirms the keyword is in the document.
  • When the results do not cover the question (no results, or only loosely related documents), follow the hint, then retry at most twice: rephrase with synonyms or the product's own terms (for example "permission control" instead of "Bucket Policy"), use grep with the product, or browse with tree. Before answering, confirm with get that the document actually answers the question; do not substitute a merely similar document. If nothing fits, say that no matching official document was found.
  • If a command fails (network error, service unavailable), tell the user that documentation retrieval is unavailable. Do not substitute model memory for official documentation or invent a source URL.
  • This Skill only covers official Baidu Cloud documentation. It does not search community articles, blogs, or other clouds' documentation.