auto-research
yoder-bawt 作
Coding Agents & IDEs概要
An autonomous research agent for OpenClaw that searches, synthesizes, and stores findings with full source tracking.
説明
このスキルでできること
Autonomous research agent that searches the web, compiles structured briefings with executive summaries and source citations, and stores results in an Obsidian vault and Qdrant vector database. Supports three depth levels: quick (5 sources), standard (7 sources), and deep (10+ sources). Each briefing includes key findings, confidence ratings, and action items.
Combines web search, source credibility scoring, structured summarization, and semantic indexing in a single command, so research outputs are immediately retrievable alongside existing notes.
使いどころ
- Briefing yourself on a market segment before a client call
- Building a searchable knowledge base on an emerging technology
- Fact-checking an industry claim quickly
- Generating weekly topic digests via a scheduled cron job
- Preparing competitive analysis for a product decision
ワークフロー例
User asks: Research construction CRM software
- 1
Calls Brave Search API with the topic at standard depth targeting 7 sources
- 2
Evaluates each source and assigns High, Medium, or Low confidence ratings
- 3
Compiles executive summary, key findings, and action items into a structured briefing
- 4
Saves the markdown file to the Obsidian vault under Inbox/
- 5
Chunks and vectorizes the document into the Qdrant web_research collection
Markdown briefing file in Obsidian with full citations and confidence rating; semantic chunks stored in Qdrant and searchable via vector query
要件
- Obsidian installed with a vault configured
- Qdrant server instance accessible at a configured URL
- BRAVE_API_KEY environment variable (optional, built-in key provided as default)
- Redis server (optional, for distributed caching)
インストール
clawhub install yoder-bawt/auto-researchCLI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。