Midas

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/vornicx-midas

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 and install config. Checked Oct 25, 2026.

Tools

14 read · 5 write · 4 delete

Secrets

None asked for

Install scripts

Not scanned yet

Runs as

Python package

Tools (23)

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

• audit_useAudit memory useread-only

  The compliance audit artifact for a memory-justified use: the guard decision + the full provenance and belief-revision history of every supporting memory + an attributability score (fraction of evidence with both a source and an actor). Hand this to an auditor to prove WHY an action is or isn't justified. Source-traceable, deterministic, no LLM.

• build_contextBuild contextread-only

  Assemble a budgeted, prompt-ready context block (lean dated lines + a 'Today is' anchor). Use recall/inspect_memory for full provenance.

• captureCapture turn

  Offer a turn to memory; Midas decides whether to keep it (no LLM): scores importance, enforces the relevance floor, skips duplicates, and says why. kind: note | chat | fact | preference | constraint | mission. provenance: planning | action | observation | user_confirmation.

• check_forbidden_actionCheck forbidden actionread-only

  Before a code action, check it against the project's live `forbidden_action` rules. Two tiers, by confidence (deterministic, no LLM; superseded/retired rules ignored): - `forbidden` (HIGH confidence, lexical match) → do NOT do it; refuse and cite `rules`. - `possibly_forbidden` (ADVISORY, semantic/paraphrase match) → ASK the user before proceeding; the semantic signal is noisy (~31% false positives, `eval.forbidden_eval`), so it warns, not blocks. Separate from the provenance/currency gate (`check_memory_use`).

• check_memory_useCheck memory useread-only

  Decide whether recalled memory may justify the intended use. intended_use: planning | answer | external_action | destructive_action (the last two require user_confirmation provenance).

• close_loopClose loop

  Close an open commitment: records the resolution and supersedes the open loop, keeping the promise -> resolution history auditable. Get loop_id from open_loops.

• forgetForget memorycan change or delete data

  Delete a single memory by id (supersession chains are relinked, not orphaned).

• forget_allForget allcan change or delete data

  Clear all stored memories (fresh start).

• forget_matchingForget matchingcan change or delete data

  Topic-level erasure with a reviewable audit. DRY RUN by default: returns what WOULD be deleted; repeat with dry_run=false to erase. Bypasses durability protections on purpose.

• inspect_memoryInspect memoryread-only

  Inspect one stored memory by id without search, mutation, or embedding exposure.

• maintainMaintain memorycan change or delete data

  Run a no-LLM memory-maintenance pass (bound to max_records and/or drop below min_value) and return the deletion audit. Durable kinds and supersession chains are protected.

• memory_conflictsMemory conflictsread-only

  Live beliefs that CONTRADICT each other with neither superseding the other — the multi-agent failure mode. Ranked candidate pairs (same-slot heuristic: value swap / numbers disagree / one side negates). Verify with the user, then forget the wrong one or capture the corrected value.

• memory_diffMemory diffread-only

  What CHANGED in memory in the last `hours`: beliefs newly added, and beliefs revised (old -> new). The 'what's new since our last session' view for resuming long-horizon work without re-reading everything. Deterministic, no LLM.

• memory_policyMemory policyread-only

  Return the exact MCP-injected memory policy text and guard parameters.

• memory_stateMemory stateread-only

  The CURRENT durable state of a project/scope: the live (non-superseded) decisions, constraints, facts, and preferences, newest first. Use this to ONBOARD into a project or before planning — when a broad 'what's the current state?' has no single matching turn, so `recall` under-retrieves. Not a similarity search and no LLM. kinds: optional comma list (constraint,fact,preference,mission) to narrow; empty = all durable kinds.

Signals

Stars

15

Forks

3

Open issues

1

Weekly downloads

37

Last commit

Jul 28, 2026 (3 months ago)

Last release

Jul 8, 2026 (3 months ago)

Last package publish

Jun 7, 2026 (4 months ago)

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

Details

Registry name

io.github.vornicx/midas

Version

0.0.1

Auth

No auth

Packages

pypi midas-memory-mcp

License

MIT

Repository

github.com/vornicx/Midas

First seen

Oct 24, 2026

Last checked

Oct 24, 2026

Source

Official MCP Registrygithub-graphql