本文へスキップ
TakoAPI
スキル一覧に戻る

wiki-retriever

urrrich 作

Search & Research
476ダウンロード数はclawskills.shが公開している値です
0閲覧数
0件のいいねサインインするとこのスキルにいいねできます
—まだ評価がありません
あなたの評価

概要

AI agent for wiki retriever tasks.

説明

このスキルでできること

Retrieves documents from a knowledge base by progressively narrowing scope: first scanning all file paths, then reading content to confirm relevance, and returning up to 10 matched files. Works with wiki task files and user-uploaded documents including Feishu files.

Reads document content before finalizing results rather than filtering on file names alone, reducing false positives in large knowledge bases.

使いどころ

  • Finding internal docs on a specific product feature
  • Locating onboarding guides across a large company wiki
  • Pulling Feishu-uploaded files related to a project
  • Answering questions by surfacing the most relevant knowledge base entries
  • Searching across mixed file sources for a specific topic

ワークフロー例

入力

User asks: Find all documents about the API authentication flow

エージェント
  1. 1

    Call get_wiki_file_paths to retrieve all file paths in the knowledge base

  2. 2

    Select files with names loosely related to API, authentication, or auth

  3. 3

    Read the content of each candidate file to confirm relevance

  4. 4

    Filter down to the most relevant matches, capped at 10

  5. 5

    Submit the final file list via submit_result with attached files

出力

A set of up to 10 confirmed documents covering the API authentication flow

インストール

clawhub install urrrich/wiki-retriever

CLIコマンドを実行するか、Claude Code、Cursor、Codexなどお使いのコーディングエージェントにプロンプトを貼り付けてください。

リンク