Search first — this is usually sufficient. Results include content, chunk IDs, and source files:
```
search_docs(query="CharacterBody3D move_and_slide")
```
Fetch if needed — if a search result's content is truncated or you need the full chunk:
```
get_chunk(chunk_id=1234)
```
Browse only when exploring a whole class — get the outline of all chunks in a file, then fetch specific ones:
```
get_doc_outline(source_file="doc/classes/CharacterBody3D.xml")
get_chunk(chunk_id=1234)
```
Related — find semantically similar chunks:
```
get_related_chunks(chunk_id=1234)
```

Setup

1. Parse the Godot docs

Clone the Godot repo and run the parser once:

cd /Users/bingles/.claude/skills/docs-reading

uv run scripts/parse_godot.py \
  --godot-repo /path/to/godot \
  --db dbs/godot.db \
  --corpus-name godot \
  --corpus-version 4.6

2. Configure the MCP server in `~/.claude.json`

Add this entry under "mcpServers":

"godot-docs": {
  "type": "stdio",
  "command": "uv",
  "args": [
    "run",
    "--python", "3.12",
    "/Users/bingles/.claude/skills/docs-reading/scripts/mcp_server.py",
    "--db", "/Users/bingles/.claude/skills/docs-reading/dbs/godot.db",
    "--corpus-name", "godot",
    "--corpus-version", "4.6"
  ],
  "env": {}
}

3. Restart Claude Code

The MCP server starts automatically on next launch. Tools (search_docs, get_chunk, etc.) will appear under the godot-docs server.

Setup

1. Parse the Godot docs

2. Configure the MCP server in ~/.claude.json

3. Restart Claude Code

2. Configure the MCP server in `~/.claude.json`

godot-docs

Typical workflow