-
Notifications
You must be signed in to change notification settings - Fork 1
Installation
# 1. Add the marketplace
/plugin marketplace add mironmax/claudecode-plugins
# 2. Install the plugin
/plugin install knowledge-graph@maxim-plugins
# 3. Disable built-in auto-memory (see below)
# 4. Install global command (optional but recommended)
bash ~/.claude/plugins/knowledge-graph/install_command.sh
# 5. Restart Claude CodeThat's it. The plugin ships hidden skills that automatically load behavioral rules into every session — no CLAUDE.md setup required.
Claude Code stores conversation transcripts in ~/.claude/projects/ — the same files that kg-scout mines to recover or build knowledge. By default it deletes sessions older than 30 days.
Add this to ~/.claude/settings.json to extend the window to 90 days:
{
"cleanupPeriodDays": 90
}This gives scout a richer history to work with and lets you recover context from older sessions when revisiting dormant projects. See Data and Backup for more detail.
Claude Code ships with a built-in auto-memory system that writes to ~/.claude/projects/*/memory/. When running alongside the knowledge graph, it creates duplicate, sometimes contradicting memory entries and wastes context window space.
Disable it:
- Open Claude Code: ⚙ Settings → Memory
- Toggle Auto-memory off
With auto-memory off, the knowledge graph becomes the single source of persistent memory — cleaner, more intentional, and without conflicts.
The plugin includes 4 hidden skills (user-invocable: false) whose descriptions are automatically loaded into Claude's context every session:
| Hidden Skill | What it carries |
|---|---|
kg-core |
Session protocol, self-awareness check, API reference |
kg-capture |
Capture triggers, search-before-put, compression rules |
kg-recall |
Proactive recall, memory traces, sync timing |
kg-maintain |
Self-reflection triggers, graph health, update discipline |
These use ~8.3K of the 16K skill description budget — well within limits, leaving room for other plugins.
By default, Claude Code will ask permission for each MCP tool call. To skip these prompts, add permissions to ~/.claude/settings.json:
{
"permissions": {
"allow": [
"mcp__plugin_knowledge-graph_kg__kg_read",
"mcp__plugin_knowledge-graph_kg__kg_put_node",
"mcp__plugin_knowledge-graph_kg__kg_put_edge",
"mcp__plugin_knowledge-graph_kg__kg_sync",
"mcp__plugin_knowledge-graph_kg__kg_search",
"mcp__plugin_knowledge-graph_kg__kg_delete_node",
"mcp__plugin_knowledge-graph_kg__kg_delete_edge",
"mcp__plugin_knowledge-graph_kg__kg_progress"
]
}
}If you already have a settings.json, merge these into your existing permissions.allow array. Don't paste the whole block — you'll get duplicate keys.
# Start the server
kg-memory start
# Check it's running
kg-memory status
# Should return JSON health response
curl http://127.0.0.1:8765/health"kg-memory: command not found"
- Run
bash ~/.claude/plugins/knowledge-graph/install_command.sh - Make sure
~/.local/binis in yourPATH - Add
export PATH="$HOME/.local/bin:$PATH"to your shell rc file
Server won't start
- Check if port 8765 is already in use:
lsof -i :8765 - Check Python venv exists:
ls ~/.claude/plugins/cache/maxim-plugins/knowledge-graph/latest/server/venv/ - Check logs:
cat /tmp/mcp_server.log
Tools not appearing in Claude Code
- Verify
.mcp.jsonpoints tohttp://127.0.0.1:8765/ - Restart Claude Code after installation
- Check server is running:
kg-memory status
Claude doesn't auto-load graph
- Check
/contextin Claude Code — hidden skills (kg-core, kg-capture, etc.) should appear - If skills are missing, the plugin may not be enabled: check
/plugin list
Claude seems confused about memory or gives inconsistent context
- Check if built-in auto-memory is still enabled: ⚙ Settings → Memory → Auto-memory (should be off)
- Python 3.10+
- Claude Code with MCP plugin support
- No external services, databases, or API keys needed