Convert all Chinese comments, docstrings, logger/print output, HTTPException detail messages, and API response messages to English across the entire codebase. Functional zh/ja localized strings (e.g. prompt templates, timezone display names, date formats) are preserved as-is. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
35 lines
887 B
Markdown
35 lines
887 B
Markdown
# rag-retrieve-only
|
|
|
|
A minimal plugin example that keeps only `rag_retrieve`.
|
|
|
|
## Overview
|
|
|
|
- Injects retrieval policy through the `PrePrompt` hook
|
|
- Exposes the `rag_retrieve` MCP server
|
|
- The plugin supports only `rag_retrieve`
|
|
- `table_rag_retrieve` is disabled
|
|
- Local file retrieval is disabled
|
|
|
|
## Directory Structure
|
|
|
|
```text
|
|
rag-retrieve-only/
|
|
├── README.md
|
|
├── .claude-plugin/
|
|
│ └── plugin.json
|
|
├── hooks/
|
|
│ ├── pre_prompt.py
|
|
│ └── retrieval-policy.md
|
|
├── rag_retrieve_server.py
|
|
└── rag_retrieve_tools.json
|
|
```
|
|
|
|
## Current Retrieval Strategy
|
|
|
|
Default order: skill-enabled knowledge retrieval tools > `rag_retrieve`
|
|
|
|
- Prefer available in-skill knowledge retrieval tools first
|
|
- Use `rag_retrieve` when they are insufficient
|
|
- Do not run multiple retrieval sources in parallel
|
|
- The plugin supports only `rag_retrieve`
|