backboard
由 chrisk60331 提供
Web & Frontend Development简介
Integrate Backboard.io for assistants, threads, memories.
描述
该技能的功能
Connects to Backboard.io through a local Flask backend to manage AI assistants, conversation threads, persistent memories, and document RAG. Supports uploading PDFs, DOCX, spreadsheets, and code files for retrieval within conversations.
Keeps assistant state, memories, and document indexes in a single local backend rather than managing them across separate services.
何时使用
- Storing user preferences that persist across chat sessions
- Uploading internal docs and querying them in conversation
- Creating purpose-specific assistants for different teams
- Reviewing conversation history across multiple threads
- Accumulating project context as memories over time
示例工作流
User asks: 'Upload my project spec and tell me what it covers'
- 1
Call backboard_upload_document with the file path
- 2
Poll backboard_document_status until processing completes
- 3
Call backboard_send_message with memory set to Auto to query the indexed document
- 4
Return the summary to the user
Document indexed and summary returned based on retrieved content
前置要求
- Backboard.io API key set as BACKBOARD_API_KEY environment variable
- Local Flask backend running on http://localhost:5100, started via ./start.sh in the backend directory
安装
clawhub install chrisk60331/backboard运行 CLI 命令,或把这段提示词粘贴到 Claude Code、Cursor、Codex 或你在用的任意编程智能体里。