技能详情(站内镜像,无评论)
版本:v1.17.0
Package:adelpro/socraticode-mcp
安装(复制给龙虾 AI)
将下方整段复制到龙虾中文库对话中,由龙虾按 SKILL.md 完成安装。
请把本段交给龙虾中文库(龙虾 AI)执行:为本机安装 OpenClaw 技能「socraticode-mcp」。简介:Agent skill by @adelpro on ClawHub.。
请 fetch 以下地址读取 SKILL.md 并按文档完成安装:https://raw.githubusercontent.com/openclaw/skills/refs/heads/main/skills/adelpro/socraticode-mcp/SKILL.md
(来源:yingzhi8.cn 技能库)
SKILL.md
---
name: socraticode-mcp
description: Install and configure SocratiCode MCP server for semantic code search and codebase indexing
metadata:
openclaw:
emoji: ""
homepage: https://github.com/giancarloerra/socraticode
requires:
bins: ["docker", "npx", "mcporter"]
env: []
claws:
title: "SocratiCode MCP - Semantic Code Search"
description: "Install and configure SocratiCode MCP server for OpenClaw. Enable semantic code search, codebase indexing, and dependency graph analysis."
tags: ["openclaw", "mcp", "socraticode", "code-search", "codebase-indexing", "semantic-search", "qdrant"]
---
# SocratiCode MCP
Install and configure SocratiCode for semantic code search and codebase intelligence on OpenClaw.
## What is SocratiCode?
MCP server providing:
- **Hybrid search** - Semantic + keyword search via RRF
- **Code indexing** - AST-aware chunking at function/class boundaries
- **Dependency graphs** - Import analysis for 18+ languages
- **Live updates** - File watcher keeps index current
## Prerequisites
| Requirement | Purpose |
|-------------|---------|
| Docker running | Qdrant vector database |
| npx (Node.js 18+) | Execute socraticode package |
| mcporter | Manage MCP servers |
## Installation
### Step 1: Install mcporter
```bash
npm install -g mcporter
```
### Step 2: Configure mcporter
Create `~/.openclaw/workspace/config/mcporter.json`:
```json
{
"mcpServers": {
"socraticode": {
"command": "npx",
"args": ["-y", "socraticode"]
}
}
}
```
Verify:
```bash
mcporter list
```
### Step 3: Start Qdrant
```bash
docker run -d --name socraticode-qdrant
-p 16333:6333 -p 16334:6334
qdrant/qdrant:v1.17.0
```
## Per-Project Indexing
To index a specific project (not the config directory), create `mcporter.json` in project root:
```bash
cd /path/to/your-project
echo '{
"mcpServers": {
"socraticode": {
"command": "npx",
"args": ["-y", "socraticode"]
}
}
}' > mcporter.json
```
Then use `--config mcporter.json`:
```bash
# Index
mcporter --config mcporter.json call socraticode.codebase_index
# Check status
mcporter --config mcporter.json call socraticode.codebase_status
# Search
mcporter --config mcporter.json call socraticode.codebase_search query="auth" limit=5
```
## Commands
| Command | Purpose |
|---------|---------|
| `codebase_index` | Start indexing |
| `codebase_status` | Check progress |
| `codebase_search` | Semantic search |
| `codebase_graph_query` | Find imports/dependents |
| `codebase_graph_visualize` | Mermaid diagram |
| `codebase_graph_circular` | Detect cycles |
| `codebase_watch` | Toggle file watcher |
| `codebase_stop` | Stop indexing |
## Troubleshooting
### Indexing Wrong Directory
Use per-project mcporter.json + `--config` flag.
### Qdrant Not Running
```bash
docker start socraticode-qdrant
# Or recreate:
docker run -d --name socraticode-qdrant -p 16333:6333 -p 16334:6334 qdrant/qdrant:v1.17.0
```
### Slow on macOS/Windows
Install native Ollama:
```bash
brew install ollama
ollama serve
```
Or use OpenAI:
```bash
export OPENAI_API_KEY=your-key
export EMBEDDING_PROVIDER=openai
```
## Agent Integration
Add to your AGENTS.md:
```markdown
## Codebase Search (SocratiCode)
- Use `codebase_search` for exploration
- Check `codebase_status` if no results
- Search before reading files
```