Myco Brain

Install

Works with Claude Code, Codex, VS Code, Cursor, Windsurf, Claude Desktop. Read the repository before you give any tool access to your data.

npx agentmarket install connector/thegoodguysla-myco-brain

About

No reviews yet. Installed it? Reviews come from verified installers only. How reviews work

Permissions

What this server can do and what it asks for, from its tool list, install config and code. Checked Oct 25, 2026.

Tools

6 read · 5 write · 0 delete · 4 not yet classified

Secrets

None asked for

Install scripts

None found

Runs as

Node.js package

Tools (15)

Read from the source code; the running server may differ. Descriptions are the server’s own words; hidden characters are shown as [U+…].

• brain_annotate

  Attach a lightweight note to the current session for continuity. Not extracted, not adjudicated, not durable workspace truth.

• brain_context_pack

  Primary read. Assemble relevant prior decisions, entities, people, and ingested documents for a task topic. Call FIRST on any task.

• brain_get_related

  Relational context and the people/entities around a topic, with provenance.

• brain_ingest

  The default write path. Hand the engine a source (a document, transcript, thread, or decision record) and it extracts entities and relations with a source and a confidence score each, runs contradiction-and-supersession, and folds them into the workspace graph. The primary way truth enters Myco; prefer it over asserting conclusions, and feed two or three independent sources when a fact matters so confidence compounds.

• brain_memory_context

  Recall durable, source-backed facts from the shared memory before you act: decisions, preferences, commitments, prices, and what changed. Ask in plain words ("What tree are we using for the Riverton plaza, and why?"): the query is ranked by how many of its distinctive words a fact matches, and by meaning when a local embedder runs, so no word has to match. Every result carries its source and its effective dates, so cite the source rather than treating recall as unstated truth. Houses are your top-level boundaries (for example your own company and each client); scope comes from this server's cr…

• brain_memory_write

  Record one durable fact the moment it appears (a decision, commitment, price, preference, or correction), with its source. Required besides the listed fields: raw_payload (the source text, as an object such as {"text": "..."}) or source.content_sha256, exactly one of the two. Facts are never silently overwritten: to change one, pass supersedes with the old assertion_id and a supersede_reason, and the history is kept. Retrying with the same idempotency_key returns the original write; reusing a key for different content is refused. The store records the fact for the instance owner with attestati…

• brain_neighbors

  Walk the knowledge graph from an entity.

• brain_propose_fact

  Submit a structured claim (subject, predicate, object) when you believe something is true but have no source to ingest. It enters a gated review queue as a candidate and is promoted when a reviewer approves it or a later ingested source corroborates it. Use brain_ingest whenever a source exists.

• brain_recall_memory

  Recall YOUR OWN private saved notes only (not workspace knowledge).

• brain_save_memory

  Your PRIVATE scratchpad, not workspace truth. Ungated, scoped to YOUR agent only, confidence hardcoded 1.0, nothing extracted, no provenance, recalled only by you. Use strictly for private working notes within a session, never for facts the workspace or other agents should trust or cite.

• brain_search

  Full-text / semantic search across ingested workspace sources.

• brain_self_check

  Agent-callable health + attention check (the self-check that talks). Returns structured signals: working (live document/chunk/embedding counts = 'it's working'), pending (the review backlog awaiting the user's approval), and problems (semantic search off, embeddings/extraction behind, approvals blocking) each with a concrete fix. Pull-only and token-cheap. Call it at the start of a session in ambient/audit mode, or whenever the user asks 'how's the brain?' or something seems off. Surface problems + pending approvals to the user; never invent numbers.

• brain_set_mode

  Set how visible Myco is (and optionally its scope). silent = invisible, ~0 tokens (default); ambient = one cheap status line when memory shaped the answer; audit = full provenance, for client/legal/financial work. Call this on a user visibility instruction: 'run silently'/'turn off stats' -> silent; 'just confirm it's working' -> ambient; 'show your sources'/'this is for a client' -> audit. Use persist:false for a one-off task ('audit THIS'); persist:true (default) saves it as the workspace default so it follows the user across clients.

• brain_stats

  Workspace memory-health snapshot.

• brain_why

  Provenance for a fact: its source chain, current confidence, and whether it is contested. Use for 'why' and 'since when'.

Signals

Stars

7

Forks

1

Open issues

0

Weekly downloads

131

Last commit

Oct 15, 2026 (13 days ago)

Last release

Oct 15, 2026 (13 days ago)

Last package publish

Oct 15, 2026 (13 days ago)

From GitHub (fetched Oct 24, 2026) and the package registry.

Details

Registry name

io.github.thegoodguysla/myco-brain

Version

1.3.0

Auth

No auth

Packages

npm @mycobrain/mcp-server

License

Apache-2.0

Repository

github.com/thegoodguysla/myco-brain

First seen

Oct 24, 2026

Last checked

Oct 24, 2026

Source

Official MCP Registrygithub-graphql