First public open-source release

Open Source, Apache-2.0

Myco Brain 1.0.0 is the first public release — a self-hosted memory layer for AI agents that gives any MCP client cross-session recall with deterministic facts and source-backed answers. The code is open source under the Apache-2.0 license at github.com/thegoodguysla/myco-brain, and the MCP server ships on npm as @mycobrain/mcp-server.

MCP Server — 11 Tools

A self-hosted MCP server exposing 11 tools to any MCP client:

• brain_context_pack — primary context assembly
• brain_search — hybrid search (vector + full-text + filters)
• brain_why — provenance chain inspection
• brain_neighbors — knowledge graph traversal
• brain_ingest — file, URL, and text ingestion
• brain_propose_fact — agent fact proposals
• brain_annotate — agent breadcrumb notes
• brain_save_memory — simplified agent memory ingestion
• brain_recall_memory — agent-scoped semantic recall
• brain_get_related — relational context query with provenance
• brain_stats — workspace memory-health snapshot

Tool contracts are stable within a major version — the inputs and outputs of the brain_* tools will not break in a 1.x release.

Search, Ingestion & Provenance

• Keyless full-text (BM25) search and document ingestion — no API keys needed to get started.
• Content-hash deduplication on ingest — re-ingesting identical content never multiplies memory.
• Queryable provenance via brain_why, with evidence summaries (“supported by N mentions across M sources”).
• Bulk ingest CLI (mycobrain-ingest) for local folders and GitHub repos.

Knowledge Graph

• Entity extraction, entity resolution, and entity-to-entity relationships.
• Runs fully local via Ollama with no API key, or with an Anthropic key for the most accurate graph.

Infrastructure

• Postgres 16 + pgvector schema with 42 versioned migrations.
• Seeded default workspace, so docker compose up works with zero configuration — one command starts Postgres, the MCP server, and the extraction worker.
• Reproducible dedup + provenance benchmark in examples/benchmark/.
• CI that boots the real Docker stack and runs all tools end-to-end.

Notes

• Vector (semantic) search requires an OpenAI key; full-text search is keyless. Local embeddings are on the roadmap.
• Managed cloud is waitlist-only and not part of this release. Self-hosting is the default.