Semantic search across memories, with optional metadata filters (author, source, date window, multiple libraries).

Browse/list memories by metadata WITHOUT a search query: inventory, not ranking. Newest first. Use to answer 'what did agent X write recently', 'everything from this source', 'all memories since a date'. Same filters as mind_search.

Unified recall across all memory types at once (facts + memories + procedures) for one query. Current facts first, then semantic memories, then matching error->fix procedures. One call instead of three separate searches.

Decide whether to search memory BEFORE answering a user query. Returns priority (must-search / should-search / answer-directly), the reason, and which libraries to search first. Call this on a turn when unsure; it implements the search-before-answer trigger policy (named project, meta-cue like 'did I tell you', negation to verify, cross-session reference). Advisory: it cannot force a search.

Open the 3D memory visualization in the user's browser: the brain as glowing cores (memories, facts, regions) wired by neurons, with live pulses. Call this when the user asks to SEE or SHOW their memory / 'the brain' / how memory looks. Spawns a local viewer and returns the URL.

Add a memory entry

Persist an externally-sourced snippet (code, doc, RFC quote, commit message, etc.) into mgi-mind with a mandatory provenance citation. The agent supplies the snippet AS PLAIN UTF-8: no HTML, no markup. Call this ONLY when the snippet was just produced by a code-search or doc-search MCP in the same session; do NOT fill provenance fields from memory.

Check system health

Generate compact context briefing for session start

Auto-ingest memory. PRIMARY: pass a candidates array of already-extracted items you judged worth keeping. BACKSTOP: omit candidates and pass raw text for marker-based heuristic extraction. Each candidate is secret-scrubbed and near-dup-checked before writing.

Record a procedural-memory lesson (error -> fix). Stored unverified by default (a manual lesson has no truth signal); surfaced with low weight until confirmed. Pass verified=true ONLY with a deterministic signal (test green / exit 0).

Recall error->fix playbooks for an error and/or a task context. Ranking blends relevance with trust: a verified or proven fix gets a boost, a repeatedly-failing one is demoted, but a strongly-matching unverified fix still surfaces.

Record whether a recalled procedure worked when reused. worked=false raises its fail count and demotes it, so the store self-corrects instead of ossifying on a bad fix.

v1.5: Record a typed external-signal outcome on any memory (not only procedures). Use this when a test passed/failed, code compiled, a user explicitly confirmed/denied a fact, or a citing memory referenced this one. Idempotent on (memory_id, signal_type, source): re-posting the same triple updates the existing entry rather than appending a duplicate. Signals contribute to the fact's external_signal_score, which feeds the duel rule and the §3 Mechanism 2 doubt-window guardrail. When memory_id is a PROCEDURE and signal_type is test_passed or code_compiled, this also bumps the procedure's success/fail counters: so a green test after a mind_learn fix marks that playbook verified without a separate mind_procedure_outcome call. Pass the procedure id returned by mind_learn.

Show recent additions chronologically

Preview what mgimind consolidate would do: count of exact duplicates, near-duplicates, and cold (old + unused) entries. Always dry-run on the MCP surface; destructive consolidation stays on the CLI where the user types --apply explicitly. Use this when the user asks 'how much duplicate memory do I have?' or before suggesting they run the CLI command.

Restore an ARCHIVED (soft-forgotten) memory by id, returning it to search. Archiving (consolidate --archive-cold) reversibly hides cold memories instead of deleting them; this is the undo.

Read a webpage as Markdown, optionally save to library

Export all data to JSON or Markdown files

Import markdown files from a directory (Obsidian, etc.)

Show memory statistics

Inspect, promote, or expire entries that the relevance gate filtered into the quarantine layer. Single tool with action: list (newest first, optional library filter), show (full content + gate reason by id), promote (the gate was too strict: move to ordinary memory by id), expire (the gate was right: delete by id; only ever touches quarantined points, never live memory, and stays recoverable from the audit log). Replaces mind_quarantine_list / _show / _promote.

Vault is terminal-only by design. This tool explains how the user runs the equivalent mgimind vault command: secret values never cross the MCP channel. Single tool with action: store (instructions for storing a secret), get (instructions for retrieving), list (instructions for listing keys). Replaces mind_vault_store / _get / _list.

Manage the agent's session record across runs. Single tool with action: start (open a new session for an agent), last (read the previous session's summary), end (close the active session with a hand-written summary). Replaces mind_session_start / _last / _end.

Knowledge-graph facts (subject -> predicate -> object). Single tool with action: add (insert a new fact), query (list facts about a subject), invalidate (soft-delete a fact by id when a single-valued fact has been superseded). Replaces mind_fact_add / _query / _invalidate.

Library namespaces. Single tool with action: create (new library by name), list (all libraries with counts), delete (remove a specific memory by id within a library: destructive). Replaces mind_create / mind_list / mind_delete.

v1.4: register and inspect predicate cardinality (single | temporal-single | multi). Cardinality controls how mind_fact(action="add") detects conflicts: Multi predicates never conflict; Single/TemporalSingle flag a conflict_pending event when a second distinct object is asserted. Default for unregistered predicates is Multi.

Pinned core-memory blocks (persona / user / current-project): a few small, named notes that are ALWAYS injected at the top of the session context, not ranked retrieval. Use for stable, always-relevant facts the agent must not have to search for. action=set overwrites, get/list read, remove deletes. Content is capped (4KB/block, 32 blocks); this is core memory, not a second store: put searchable knowledge in mind_add instead.

Skills: the house way of doing a kind of work, matched against the task you are ABOUT TO START (mind_recall is the other half: it answers an error that already happened). action=match takes a task description and returns the playbooks that apply, ranked by relevance and by how they have actually worked out. Call it before starting non-trivial work, the way you would check for a house rule. action=set writes one (name is the identity, so writing the same name edits it and keeps its history), list/show read, remove deletes, outcome records whether applying it worked (verify=true only for a deterministic signal, never a hunch).

DEPRECATED: use mind_fact(action="add"). Removed in v2.0. (Add a knowledge graph fact)

DEPRECATED: use mind_fact(action="query"). Removed in v2.0. (Query facts about a subject)

DEPRECATED: use mind_fact(action="invalidate"). Removed in v2.0. (Soft-delete a fact by ID (for superseding a changed single-valued fact))

DEPRECATED: use mind_session(action="start"). Removed in v2.0. (Start a new session)

DEPRECATED: use mind_session(action="last"). Removed in v2.0. (Get last session summary)

DEPRECATED: use mind_session(action="end"). Removed in v2.0. (End the active session for an agent)

DEPRECATED: use mind_library(action="create"). Removed in v2.0. (Create a new library)

DEPRECATED: use mind_library(action="list"). Removed in v2.0. (List all libraries)

DEPRECATED: use mind_library(action="delete"). Removed in v2.0. (Delete a specific memory by ID)

DEPRECATED: use mind_quarantine(action="list"). Removed in v2.0. (List entries the v0.11 relevance gate filtered into the quarantine layer. These are not surfaced by mind_search by design: use this tool when you suspect a fact was filtered (e.g., the user keeps repeating something the gate would reject as low-signal). Newest first.)

DEPRECATED: use mind_quarantine(action="show"). Removed in v2.0. (Show a single quarantined entry with its full content and the gate reason. Returns 'not in quarantine' for regular memory ids: the surface is honest about what it can see.)

DEPRECATED: use mind_quarantine(action="promote"). Removed in v2.0. (Explicitly promote a quarantined entry to ordinary memory by id. The automatic promotion path is re-asserting the same content via mind_ingest; use this when you know the entry should be live without re-ingesting.)

DEPRECATED: use mind_vault(action="store"). Removed in v2.0. (Explain how to store a secret (never accepts the secret over MCP))

DEPRECATED: use mind_vault(action="get"). Removed in v2.0. (Explain how to retrieve a secret (never returns plaintext over MCP))

DEPRECATED: use mind_vault(action="list"). Removed in v2.0. (Explain how to list stored secret keys (terminal-only; needs the master password))