| name | logvalet:document-search |
| description | Search Backlog documents by keyword within a space: runs lv document search and presents results as a snippet-based digest — title, project, and a relevant excerpt from each document. TRIGGER when: user says "ドキュメント検索", "document search", "資料を探して", "ドキュメントを探して", "Backlog ドキュメント検索", "wiki 検索", "ドキュメントを検索", "document を検索", "search documents", "find documents", "資料検索", "ドキュメントから探して", "document を探したい", "ドキュメントで調べて", "search backlog docs", "find in documents", "ドキュメント一覧から探す", "キーワードでドキュメントを検索", "keyword document search", "ドキュメントの中を検索", "資料を見つけて". DO NOT TRIGGER when: user wants to get a specific document by ID (use lv document get), wants to list all documents in a project (use lv document list), or wants to search issues/wiki (use the respective commands).
|
logvalet:document-search
Backlog ドキュメントをキーワードで横断検索し、スニペット付き digest を返す。
For full logvalet CLI documentation, see the logvalet skill.
When to use this skill
Use logvalet:document-search when you need to:
- find Backlog documents containing a specific keyword
- locate design documents, meeting notes, or knowledge base articles
- search across all projects in a space for relevant documentation
- answer "is there a document about X?" questions
Workflow
Step 1: Identify the keyword
Extract the search keyword from the user's request. If multiple keywords are mentioned, combine them or use the most specific one.
Optionally identify project key(s) if the user wants to narrow the search to specific projects.
Step 2: Search documents
lv document search "<keyword>" [--project KEY ...] [--detail snippet|meta|full] [-f json]
| Option | Default | Description |
|---|
<keyword> | required | Search term |
--project KEY | (all projects) | Filter by project key (can repeat) |
--detail | snippet | snippet: excerpt only / meta: title+metadata / full: full body |
--count | 100 | Max results (1-100) |
--offset | 0 | Pagination offset |
--sort | (API default) | created | updated |
--order | (API default) | asc | desc |
Default behaviour: searches all projects within the current space. Pass --project to narrow.
Step 3: Present results
Present the digest to the user:
- Show
total_returned and note if possibly_more=true (more results may exist)
- For each document: title, project, and the
snippet excerpt
- If
possibly_more=true, suggest using --offset to paginate
Example output format:
Found 5 document(s) matching "OAuth". (possibly more — use --offset to paginate)
1. [認証フロー設計書] (PROJ) — "...OAuth 2.0 の認可コードフローを採用。クライアント ID は..."
2. [OAuth 実装メモ] (INFRA) — "...BacklogのOAuthエンドポイント: /oauth2/token を使用..."
Step 4: Handle edge cases
- No results: Suggest broadening the keyword or removing project filters
- possibly_more=true: Offer to fetch the next page with
--offset N
- detail=full requested: Warn that full body may be large; recommend
snippet first
MCP tool (alternative to CLI)
logvalet_document_search
keyword: string (required)
project_keys: string[] (optional)
sort: string (optional) — created | updated
order: string (optional) — asc | desc
count: number (optional, default 100)
offset: number (optional, default 0)
detail: string (optional, default "snippet") — snippet | meta | full
Tips
- Cross-project search: Omit
--project to search all projects in the current space (default)
- Pagination: If
possibly_more=true, use the next_offset value from the response as --offset N to get the next page. Example: if next_offset=50, run lv document search "keyword" --count 50 --offset 50
- Keyword scope: Backlog searches document content (title and body). Use specific technical terms for best results
- Multi-space: When multiple Backlog spaces are connected, results are per-space (fan-out). Each space returns its own digest
