Setup & Installation
Or with the ClawHub CLI, for registry-managed skill folders outside a full OpenClaw workspace:
Version History
- Added explicit OpenClaw-compatible metadata under a new metadata key in SKILL.md. - Metadata specifies required binaries, environment variables, Python dependencies, and credential instructions. - No functional or workflow changes.
What This Skill Does
Connects an AI agent to Ragie.ai to perform retrieval-augmented generation. Handles the full lifecycle: ingesting files or URLs, polling until documents are ready, retrieving relevant chunks, and constructing grounded answers that cite source documents.
Ragie offloads chunking, embedding, vector indexing, and optional reranking so the agent only needs to handle ingestion triggers and prompt construction.
When to use it
- Ask questions about uploaded policy or compliance documents
- Ingest a PDF and query its contents without manual parsing
- Check which documents are indexed and delete outdated ones
- Answer support questions grounded in a product knowledge base
- Retrieve context from partitioned tenant data in a multi-user system
Example Workflow
Here's how your AI assistant might use this skill in practice.
User asks: 'What does our refund policy say about digital products?'
- 1Run retrieve.py with the user question and --rerank flag
- 2Receive scored chunks with document_name and text fields
- 3Concatenate chunk text into a grounded system prompt
- 4Generate answer using only the retrieved context
- 5Cite the document_name for each factual claim in the response
A grounded answer referencing the specific document, or an explicit statement that the information is not in the knowledge base
Requirements
Accounts, API keys, or tools you or your AI assistant may need to set up while using this skill.