
World Model Mcp
A temporal knowledge graph that sits between your coding agent and your codebase to enforce constraints and prevent regressions. It validates API references before the agent calls them, learns from your corrections across sessions, and re-injects top constraints after context window compaction. The 26 MCP tools cover constraint management, contradiction resolution with confidence weighting, and compaction audit logs. Runs over stdio or HTTP, works with Claude Code, Cursor, and pi. The setup command auto-seeds the graph from your existing code, and the AGENTS.md reader pulls conventions from declarative markdown files alongside the SQLite store. Reach for this when you want long-term memory that survives context resets and stops your agent from repeating the same hallucinated function calls.
- GitHub stars
- 24
- tools
- 8
- auth
- Secret needed
- transport
- STDIO
- env vars
- 4
- last updated
- Jul 24, 2026
Tools
8 toolsPublic tool metadata for what this MCP can expose to an agent.
query_factFetch a stored fact with provenance + confidence + source
Fetch a stored fact with provenance + confidence + source
No parameter schema in public metadata yet.
record_eventAppend an event to the audit chain (`event_type` from a fixed enum)
Append an event to the audit chain (`event_type` from a fixed enum)
No parameter schema in public metadata yet.
validate_changeCheck a code-change proposal against learned constraints, returns violations + suggestions
Check a code-change proposal against learned constraints, returns violations + suggestions
No parameter schema in public metadata yet.
get_constraintsList all constraints matching an entity or file pattern
List all constraints matching an entity or file pattern
No parameter schema in public metadata yet.
record_correctionStore a user correction so it re-surfaces on next relevant query
Store a user correction so it re-surfaces on next relevant query
No parameter schema in public metadata yet.
get_related_bugsFetch bugs touching the given files + a per-file risk score
Fetch bugs touching the given files + a per-file risk score
No parameter schema in public metadata yet.
seed_projectBulk-ingest an existing codebase into the knowledge graph
Bulk-ingest an existing codebase into the knowledge graph
No parameter schema in public metadata yet.
ingest_pr_reviewsTurn recent PR review comments into constraints
Turn recent PR review comments into constraints
No parameter schema in public metadata yet.
Sponsors
Configuration
WORLD_MODEL_DB_PATHdefault: ./.claude/world-model/Database location
ANTHROPIC_API_KEYsecretOptional - enables LLM-powered entity extraction
WORLD_MODEL_EXTRACTION_MODELdefault: claude-3-haiku-20240307Model for entity extraction
WORLD_MODEL_REASONING_MODELdefault: claude-3-5-sonnet-20241022Model for complex reasoning
Readme
World Model MCP
Persistent memory + optional signed audit for AI coding agents.
world-model-mcp is persistent memory plus a post-quantum-signed, offline-verifiable audit trail for AI coding agents (FIPS 205 hybrid Ed25519 + SLH-DSA), MIT-licensed and fully local.
world-model-mcp ships a local SQLite knowledge graph your agent queries every turn: hallucinations become verifiable, corrections stick across sessions, and regressions get caught before they land. Flip on the audit chain and every event is signed with FIPS 205 hybrid Ed25519 + SLH-DSA, verifiable offline forever. MIT-licensed, runs entirely local, works with 10+ AI coding agents including Claude Code, Cursor, Codex, Continue, Cline, Windsurf, GitHub Copilot Chat, pi, OpenClaw, and Hermes Agent.
Details
- Repository
- github.com/saravananjaichandar/world-model-mcp
- Categories
- Documents & Knowledge, Automation & Workflows
- Package
world-model-mcp- Transport
- STDIO
- Auth
- Required
- Registry
- MCP registry, active