Graphiti

Graphiti — Temporal Knowledge Graph Memory

Graphiti provides persistent AI memory through a temporal knowledge graph backed by Neo4j. Store and retrieve project context, solutions, user preferences, and architectural decisions across sessions.

Decision Tree

Starting work?          → search_nodes for project context
Looking for patterns?   → search_memory_facts for relations
Solved hard problem?    → add_memory with the solution
Need recent context?    → get_episodes for session history
Session ending?         → add_memory with key learnings
User shares preference? → add_memory to remember it

MCP Tools

Core Workflows

Session Start (ALWAYS)

search_nodes({ query: "[project-name] context" })
search_memory_facts({ query: "[project-name] patterns decisions" })
get_episodes({ max_episodes: 5 })

Problem Solving

1. search_memory_facts({ query: "[error message or problem]" })
2. If found → Apply known solution
3. If not → Debug normally
4. If solved after >15 min effort → add_memory with solution (see references/store-patterns.md)

Storing Memories

For the four detailed templates (Failure→Success, Infrastructure Context, Tool Config, Cross-Session Context) including the session-summary template, see references/store-patterns.md.

Minimal pattern:

add_memory({
  name: "Solution: [brief title]",
  episode_body: "Problem: ... Context: ... Solution: ... Why: ...",
  source: "text",
  source_description: "debugging session",
  group_id: "[project-name]"
})

Auto-Store Triggers

Store automatically (without being asked) when ANY of these occur:

When to Add Memory

DO Add:

• Solutions to non-trivial problems (>15 min debugging)
• Failed approaches and what worked instead
• Infrastructure topology and service relationships
• User-explained project context or preferences
• Architectural or design decisions with rationale
• Discovered undocumented behavior or gotchas
• Session summaries with key learnings
• Tool configurations with gotchas

DON'T Add:

• Generic knowledge (already in training data)
• Trivial fixes (typos, simple syntax errors)
• Sensitive data (credentials, API keys, PII, connection passwords)
• Temporary debugging info
• Content that changes frequently
• Exact secret values (store the 1Password path instead)

Group IDs

Use group_id to scope memories per project or domain:

add_memory({ name: "...", episode_body: "...", group_id: "nix-dotfiles" })
search_nodes({ query: "...", group_ids: ["nix-dotfiles"] })

Entity Types

Graphiti auto-extracts: Preference (user choices), Requirement (features/constraints), Procedure (step-by-step processes), Organization (companies/teams), Document (files/references), Topic (knowledge domains).

Search Tips

# Find architecture context
search_nodes({ query: "nix-dotfiles architecture structure" })

# Find solutions to errors
search_memory_facts({ query: "error ENOENT file not found" })

# Find user preferences
search_nodes({ query: "user preference coding style" })

# Pagination
search_nodes({ query: "...", max_nodes: 20 })

Server Setup

Backend: Neo4j (graph DB via launchd) MCP Server: Graphiti MCP on http://127.0.0.1:51847

Both services run as macOS LaunchAgents (auto-start on login):

Check / restart:

launchctl list | grep claude
launchctl kickstart -k gui/$(id -u)/com.claude.neo4j
launchctl kickstart -k gui/$(id -u)/com.claude.graphiti

Shell CLI (Without MCP)

For hooks, CI, and shell automation, use the bundled scripts in scripts/. See references/shell-cli.md for the full CLI reference, examples, and HTTP tool-name mapping.

Troubleshooting