MGI-Mind

MGI-Mind

English | Русский | 中文

Latest release: v2.8.2 · CHANGELOG · Discussions · Issues · Contributing

[MGI-Mind MCP server]

Local long-term memory for AI assistants. One Rust binary, a local Qdrant vector database, local ONNX models. Speaks MCP, so Claude Code and other assistants read and write memory on their own. Also a normal CLI.

[Store a fact, then recall it by meaning: a query with no shared keywords still finds it]
Store a fact in plain words, recall it with a query that shares none of them. Semantic search matches intent. Your assistant runs the same mind_search on its own over MCP. Regenerate with vhs docs/demo.tape.

The same exchange inside an assistant:

You:  what was the deploy server again?
Assistant (calls mind_search "deploy server"):  -> "Deploy server is 10.0.0.5:8080, SSH as deploy@, key in vault"  (source: infra.md)
You:  right, thanks

Nothing leaves the box. Embeddings, search, reranking and the vault are all local. No cloud account, no API key, no telemetry.

Contents

• What it is
  • The validity model
• Why bother
• Quick start
• Using it
• How it works
• Command reference
• Configuration
• Languages and the reranker
• Changing the embedding model
• Troubleshooting
• Security
• Status and audit
• Project layout
• License

What it is

MGI-Mind sits between you and your assistant. The assistant writes short notes ("memories") and facts as a conversation goes, and pulls the relevant ones back when they matter.

Retrieval, not just storage:

• Hybrid search. Each memory is stored as two vectors: a dense vector (multilingual-e5-base, 768 dims) for meaning, and a sparse term-frequency vector (TF-IDF, BM25-style) for exact words. A query runs both arms and fuses them with Reciprocal Rank Fusion. "The server box" finds "deploy host" through the dense arm; fossilize_replay finds the one note that contains that exact token through the sparse arm.
• Cross-encoder reranking. Fused top candidates are re-scored by bge-reranker-base, which reads the query and each passage together and is more accurate than comparing vectors. On by default and English-tuned; see Languages and the reranker for the trade-off on other languages.
• One warm process. mgimind mcp is the MCP server itself: it runs for the whole session, models load once and stay warm in memory, so a lookup costs milliseconds instead of reloading on every call.

Around retrieval there's a knowledge graph for structured facts, per-agent session logs for cross-session continuity, and an encrypted terminal-only vault for secrets.

The validity model

Hybrid search is table stakes; most memory tools have some version of it. The part that is harder to find is what keeps the store from ossifying as it fills with old, contradictory, or self-reinforcing beliefs. These run mostly on their own:

• Duel rule. When a new fact contradicts an existing one on the same subject and predicate, the second write resolves against the first instead of piling up a second "truth". An entrenched fact (many dependants, confirmations, age) holds; a strong fresh fact flips it and dampens the loser to a hidden stale status; a borderline one is marked contested or diverted to quarantine. Confirmations accrue live: re-asserting the same triple raises its own confirmations_count, so a fact confirmed a few times resists a single fresh contradiction (contested, not flipped) while an unconfirmed young fact still flips as easily as before. dependants still only comes from the migrate-v14 dependants --apply backfill - a cheap live producer (it would mean walking the graph on every write) is not implemented. Nothing is deleted, so the audit log keeps the loser; a re-proposed stale loser gets a fair new duel against whatever is current, keeping its own original created_at and confirmation history rather than starting over. Automatic, inside a normal mind_fact add.
• Doubt window. An entrenched fact has to keep re-justifying itself. Every fact query measures how far the current context has drifted from the context the fact was written in; after five consecutive drifted retrievals the fact stops outranking facts that still hold up, and any single in-context retrieval resets the count. Nothing is hidden or deleted. Drift is measured on vectors with the corpus mean removed, because raw cosine cannot separate topics in an anisotropic embedding space (see mgimind calibrate below). The current context is the running process's own recent queries, so this applies inside a warm mgimind mcp or serve-http session; a one-shot CLI command has no context yet and no fact is penalised for that. This ships in shadow: drift is measured and recorded, and no fact changes, until you set doubt_drift_threshold from your own store's numbers and switch doubt_drift_mode to "enforce". A background pass separately re-tests entrenched facts that have gone quiet, under hard guarantees: never during a tool call, a per-tick cap, and a load-aware cadence.
• Inheritance discount. mind_context's briefing (what a session sees at start) marks every fact it surfaces as inherited for that process. If a duel challenger is a re-proposal of one of those inherited facts - the agent echoing back what it was told rather than independently reconfirming it live - that challenge counts at half weight. The flag clears the moment the agent actually re-asserts the fact live and it lands without a conflict. Single-source decay (repeated confirmation from the same source counting for less each time) has the formula (diversity_weighted_count) but no live caller yet - not implemented, no "cannot co-confirm" guarantee today.
• Two time axes on facts. Record-time (created_at/valid_until: when the store learned/retired a value) is bi-temporal with the superseded chain: a predicate can be registered (mind_predicate) as single-valued, temporal-single (one current value, but the previous ones are kept and queryable by date via mind_history), or multi-valued, and a superseded value is hidden from default ranking, not erased. An optional second axis, world-time (valid_from, settable on add), records when a fact became true in the world instead of when the store heard about it; mind_fact action=query with as_of + world_time=true resolves point-in-time queries against it, falling back to record-time per fact when it has no valid_from. Existing facts need no migration - they simply have no valid_from and behave exactly as before.
• Explicit invalidate is final. mind_fact action=invalidate is a one-way door: re-adding the same triple afterward refuses (verdict invalidated) instead of silently reviving it. mind_fact action=restore is the only way back.
• Typed outcome signals. mind_outcome records that a remembered fix actually worked (test_passed, code_compiled, user_confirmed, cited_by). A real success raises a memory's weight and marks a procedure verified; a failure pulls the weight down rather than being ignored.

These are research-shaped mechanisms, and their tuning constants are still being calibrated (the TODO(phase-4-calibration) markers in duel.rs / doubt.rs are honest). The shape is the point: a memory that argues with itself and demotes what stops holding up, instead of accreting forever.

mgimind calibrate measures whether that shape behaves as intended. It runs a corpus of realistic conflict situations through the live duel formulas and reports how many land on the outcome a person would expect: a fresh unsupported claim cannot overturn an entrenched belief, a CI signal can, repetition alone coexists rather than overwriting. Today 14 of 15 scenarios match intent; the one that does not is printed with its reason, not hidden, and the suite runs in CI. It also reports the doubt window's drift distribution, and that report exists because of what it found. The drift threshold was 0.4, picked in 384-dim MiniLM space and left in place when the default model became 768-dim multilingual-e5. Measured on a live store, 400 memories and all 79800 pairs, raw cosine ran from 0.666 to 0.972: the threshold could not fire once. The corpus mean vector had norm 0.884, so almost all of that cosine was a component every embedding shares. Centering fixes the measurement, and calibrate proposes a threshold from the drift your own store produces rather than from a constant that travelled between embedding spaces.

This measures the shape of the validity model, separate from retrieval recall (R@k, see BENCHMARKS.md). The two are different numbers and never belong in the same table.

Why bother

An assistant without memory asks for the same context every session and can't build on yesterday's work. The usual workaround is you keeping notes, tags, and folders, which the assistant still can't read by meaning.

The thing MGI-Mind does that Obsidian and Notion don't: the system decides what to write down, you don't. The MCP server reads what the assistant is doing in real time and routes facts, decisions, and fixes into the store through a relevance gate. You don't file, you don't tag, you don't decide what's worth saving. Low-signal candidates land in a quarantine layer (recoverable on re-assertion) instead of polluting retrieval.

How it compares to the obvious alternatives:

• Plain notes (Obsidian, Notion). Strong as your personal notebook, but the assistant can't search them by meaning, and every keystroke is equally important to a folder of .md files.
• Bare vector database. Semantic search, but no exact-term matching, no reranking, no dedup, no sessions, no facts, no secrets handling, no relevance gate, no procedural memory. You assemble all of that.
• Hosted memory API. Your data on someone else's servers. Closes the box on inspection, dedup behaviour, and what the relevance gate is actually doing.

MGI-Mind is the assembled local version: hybrid + reranked retrieval, the relevance gate, dedup, facts, sessions, procedural memory ("error → fix" playbooks), and a terminal-only vault, all behind one binary you run yourself.

A note on evaluating it. BENCHMARKS.md reports retrieval recall (R@k on LongMemEval-S, zero-API, no LLM judge): given a question, did the gold session land in the top-k. On the default CPU config that is R@5 ≈ 98%. This is not the same metric as the QA-accuracy numbers other memory tools publish (they run an answerer and a judge over their retrieval), so do not put it in the same table. It measures whether the evidence was retrievable, not whether an LLM then answered correctly. mgi-mind has no answering step, so retrieval recall is the number it actually owns.

The honest gap: the validity model (duel rule, doubt window, supersession) is the differentiator, and retrieval recall does not test it. Two things test it instead. mgimind calibrate checks behavioral shape against intent (14/15 today, run it yourself); it proves the mechanism does what a reader expects, not that the constants are tuned against real data. The external belief-revision benchmark is STALE, and there only a preliminary partial run exists (N=155, ~32%, raw verdicts under benchmark/results/), produced by the harness on a branch rather than the scaffold on main, on a reduced haystack, possibly while the duel rule was still broken. See BENCHMARKS.md for the full caveats. Treat the validity model as a designed mechanism whose shape is verified and whose constants are not yet tuned, not a proven win on external data.

Quick start

Docker (the whole brain in one command)

The image bakes the binary, a bundled Qdrant, the ONNX runtime, and the models, so it runs offline with nothing to install:

docker run -p 8765:8765 madgodinc/mgi-mind

It prints a bearer token on startup. Point the Python client or any agent at http://localhost:8765 with that token. Persist memory across restarts with a volume, and pin a stable token:

docker run -p 8765:8765 -e MGIMIND_TOKEN=your-token \  -v mgimind-data:/data madgodinc/mgi-mind

Plug it into an agent

Give an agent two tools, recall and remember, and its memory survives the process. With the server running and the Python client installed:

from mgimind import Memory
mem = Memory()                         # MGIMIND_URL / MGIMIND_TOKEN from the envmem.add("Prod DB is Postgres 16 on db-prod:5432")for hit in mem.search("prod database"):  # structured results: id, score, author    print(hit["score"], hit["content"])

That wrapping is the whole integration. Copy-paste examples for LangGraph, Pydantic AI, the OpenAI Agents SDK, and raw HTTP (any language) are in docs/integrations. MCP and stdio still work for editor clients; the HTTP surface is the door for agent frameworks.

npm (any MCP client)

With Node.js 18 or newer, the npm package installs the prebuilt binary for your platform, checked against the release's SHA-256:

npx -y mgi-mind init          # creates ~/mgimindnpx -y mgi-mind doctor --fix  # downloads Qdrant and the modelsclaude mcp add mgimind -- npx -y mgi-mind mcp

Other clients take the same command in their MCP config: "command": "npx", "args": ["-y", "mgi-mind", "mcp"].

Native install

One command. The installer drops the binary on PATH and runs init + doctor --fix (which pulls Qdrant, ONNX Runtime, and the models). Or just tell your coding agent to install mgi-mind using AI_INSTRUCTIONS.md: it covers the install itself, not only how to use it once installed.

Linux / macOS:

curl -fsSL https://raw.githubusercontent.com/madgodinc/mgi-mind/main/install.sh | sh

Windows (PowerShell):

irm https://raw.githubusercontent.com/madgodinc/mgi-mind/main/install.ps1 | iex

When it finishes, the printed command wires the server into Claude Code:

claude mcp add mgimind -- /home/you/.local/bin/mgimind mcp

mgimind mcp IS the MCP server; it runs for the whole session with the models warm and brings up the bundled Qdrant on first use. Point your assistant at AI_INSTRUCTIONS.md once so it knows the protocol (log a session, search before answering, use the vault for secrets).

doctor --fix downloads the models into ~/mgimind/: the embedder (multilingual-e5-base, quantized ONNX, ~270 MB) and the reranker (bge-reranker-base, quantized ONNX, ~280 MB). ONNX Runtime and the Qdrant binary land next to the mgimind binary itself, so a doctor that reports a missing library is telling you about <install dir>/, not ~/mgimind/.

Installer flags

• INSTALL_DIR=/opt/mgimind curl ... | sh: install somewhere other than ~/.local/bin.
• MGIMIND_TAG=v2.8.2 curl ... | sh: pin a specific release instead of latest.
• SKIP_DOCTOR=1 curl ... | sh: just drop the binary; run init + doctor --fix yourself later.

Manual install (no installer)

If you'd rather not pipe a script to a shell, grab the release tarball for your OS from Releases, put mgimind on your PATH, then:

mgimind initmgimind doctor --fixclaude mcp add mgimind -- /absolute/path/to/mgimind mcp

Try it from the CLI (the same binary works without an assistant):

mgimind create workmgimind add work "Deploy server is 10.0.0.5:8080, SSH as deploy@"mgimind search "how do I reach the deploy box"

Per-OS notes

• Linux x86_64, macOS arm64 and x86_64, Windows x86_64: prebuilt binaries in every release. The installer picks the right one.
• macOS PATH: zsh does not include ~/.local/bin, so the installer appends the export PATH line to your ~/.zshrc and tells you it did. Open a new terminal before typing mgimind. Set MGIMIND_NO_PROFILE=1 to keep the installer out of your dotfiles and add the line yourself.
• macOS Intel and ONNX Runtime: 1.24 dropped osx-x86_64, so Intel Macs get ONNX Runtime 1.23.0 while everything else gets 1.24.2. The binary requests C API 23, which both runtimes serve.
• macOS first-run quarantine: the installer clears it, but a binary downloaded through a browser needs xattr -d com.apple.quarantine /path/to/mgimind. Releases are ad-hoc signed, not notarized; on Sequoia and later the right-click → Open trick no longer works for command-line binaries, so use xattr, or approve it once under System Settings → Privacy & Security.
• Windows: SmartScreen may warn on the unsigned mgimind.exe ("Windows protected your PC"); choose More info → Run anyway. Antivirus can also quarantine the binary or the models it downloads; if mgimind doctor reports a file as downloaded but missing, allow mgimind.exe and the %USERPROFILE%\mgimind folder in your AV, then re-run mgimind doctor --fix. Code signing to remove the SmartScreen prompt is on the roadmap.

Build from source

Rust toolchain (rustup); no other dependencies.

git clone https://github.com/madgodinc/mgi-mind.gitcd mgi-mindcargo build --release                  # binary: target/release/mgimind

Then run target/release/mgimind init && target/release/mgimind doctor --fix and wire it in with claude mcp add mgimind -- /absolute/path/to/target/release/mgimind mcp.

Using it

By hand (CLI):

mgimind add notes "Mom's birthday is March 14, she likes peonies" --source personalmgimind search "when is my mother's birthday"#  1. [notes] (score: 0.94) Mom's birthday is March 14, she likes peonies#     source: personal
mgimind fact add "user" "prefers" "Rust"mgimind fact query "user"#   user -> prefers -> Rust
mgimind history --limit 3              # the three most recent memoriesmgimind stats                          # counts per library, facts, sessions

Duel rule on contradictions. When two facts contradict and the predicate is Single or TemporalSingle, the second add resolves against the first instead of piling up:

# Tell the system this predicate has one current value at a time:echo '{"jsonrpc":"2.0","id":1,"method":"tools/call",  "params":{"name":"mind_predicate","arguments":    {"action":"register","predicate":"lives_in",     "cardinality":"TemporalSingle"}}}' \  | mgimind mcp
mgimind fact add "Alice" "lives_in" "Prague"mgimind fact add "Alice" "lives_in" "Dublin"mgimind fact query "Alice"#   Alice -> lives_in -> Dublin# (Prague is preserved as history, queryable via mind_history / audit log#  but hidden from default ranking.)

If you migrated from a pre-v1.7 install where the duel rule wasn't firing at the read path, run mgimind migrate-v14 redo-duels --apply once to collapse legacy contradictions to canonical answers. The walk is idempotent and dry-run by default.

Through your assistant (MCP). Once connected, you just talk:

You:  remember that the staging DB password is in the vault under "staging-db"Assistant: (mind_add) saved. The secret itself stays in your terminal vault, not here.
You:  what database are we using on staging?Assistant: (mind_search "staging database") Postgres 16, host db-staging.internal:5432

Search returns results in tiers so the assistant spends tokens carefully: --tier 1 is a ~100-character snippet, --tier 2 (default) is ~500, --tier 3 is the full text.

Seeing the store

mgimind brain renders the memory as a graph: cores for memories, facts, and regions, wired by the links between them, pulsing as the store changes. It binds to loopback only and needs the token printed alongside its URL.

Treat it as a control surface. An unrestricted viewer also serves DELETE /api/memories/:id, PATCH /api/node/:id and quarantine promotion, so whatever reaches that port with that token can change the store. For a viewer that cannot write, start it confined to an allowlist:

mgimind viewer --libraries work,personal

That fail-closes every mutating route with a 403, along with the routes that span all libraries (graph, audit, pulse, the ingest feed). mgimind brain takes no flags, so it is always the unrestricted form.

[Memory rendered as a brain, glowing cores wired by neurons]

How it works

  your AI assistant        |  MCP (JSON-RPC over stdio)        v  mgimind mcp  (one Rust process: MCP server + embedder, models stay warm)        |  starts on first use        v  Qdrant (local, loopback only)        |        one "memories" collection, two vectors per point:        dense (e5, meaning) + sparse (TF-IDF, exact terms)

Storage. All memories live in one Qdrant collection. A library field on each point separates namespaces (work, personal, a project), and a query can filter to one library or search across all. A point's ID is a UUIDv5 of library + content, so adding the same text twice overwrites the same point, with no duplicates and no race. A created_at datetime index lets history return the newest N directly without scanning.

Embeddings. Text is embedded locally through ONNX Runtime. Default is multilingual-e5-base (768 dimensions), strong on English and handles mixed languages. The embedder is model-aware: pooling (mean or CLS), token_type_ids input, query/passage prefixes are all config, so switching models doesn't need a code change. Inputs cap at 512 tokens; add splits long text into chunks so nothing past the cap silently disappears.

Search. The query is embedded once. Qdrant runs a dense nearest- neighbor search and a sparse search in a single Query API call and fuses them with RRF. If reranking is on, the top rerank_top_k candidates are re-scored by the cross-encoder and reordered. A library filter, when given, applies to both arms.

Safety. Downloads check against pinned SHA-256 hashes (fail-closed). Qdrant binds to loopback only and can require an API key. The vault is terminal-only, and the master password and decrypted secrets never travel over the MCP channel. File writes are atomic (temp file, fsync, rename, fsync the directory), so a crash leaves the old file or the new one, never a corrupt one.

Command reference

Memory

Libraries

Knowledge graph

Skills

A procedure answers "this broke, what fixed it last time". A skill answers "I am about to do X, is there a house way to do X". Same storage and the same outcome signals, matched at a different moment: against the task, before the work. A context render carries the catalogue (names and triggers), so an assistant knows what exists without searching, and pulls the body only for the skill that applies.

Sessions

Vault (terminal only)

Service and data

Configuration

Config lives at ~/mgimind/config.json. Fields that affect retrieval:

Languages and the reranker

The default stack is tuned for English, because that's where the assistant itself reasons best and where the models are strongest. It's the recommended setup for an English-first project.

A few honest details:

• The embedder (multilingual-e5-base) is genuinely multilingual. An English query finds a Russian note and vice versa, and search alone works well across languages.
• The default reranker (bge-reranker-base) is English-tuned. It improves English ranking, but it lowers ranking quality on Russian (and other non-English languages). This is the one place the defaults favor English.
• If your content is mostly Russian (or another non-English language): set rerank_enabled = false. Hybrid dense+sparse search on its own ranks those languages well; the English-tuned reranker is what hurts them. Or swap in a stronger multilingual reranker.
• Reranking costs cross-encoder inference per query: roughly one to two seconds for 20 candidates on a CPU-only box. Lower rerank_top_k or turn reranking off for snappier search.

Changing the embedding model

Switching models usually changes the vector dimension, so existing memories must be re-embedded:

1. Back up first: mgimind backup ~/mgi-backup.tar.gz.
2. Set model_name, vector_size, pooling, uses_token_type_ids, and the prefixes in config.json for the new model.
3. mgimind doctor --fix to download it. Only the bundled defaults are pinned; a custom model downloads with an "integrity not verified" warning, so pin its SHA-256 in integrity.rs if you want strict verification.
4. mgimind migrate to re-embed everything from the stored text under the new model.

Troubleshooting

• "Model not found ... run doctor --fix": the model is not in ~/mgimind/models/. Run mgimind doctor --fix.
• "invalid expand shape" / inference errors: usually an input far over 512 tokens. add chunks automatically; if you call the library directly, chunk first.
• Searches are slow: that's the reranker on CPU. Lower rerank_top_k, or set rerank_enabled = false. Models stay warm for the life of the mgimind mcp process, so only the first lookup of a session pays the load cost.
• A tool fails right after install: run mgimind doctor (the assistant can call mind_doctor); it reports exactly what's missing (Qdrant not running, a model not downloaded, ONNX Runtime absent, a file quarantined by AV) and --fix downloads what it can.
• Dimension mismatch warning: a collection's vectors don't match vector_size, usually after a model change. Re-embed with mgimind migrate.
• Russian results feel off: set rerank_enabled = false (see the language note above).
• A VPN, tunnel or proxy breaks while mgimind is running: run mgimind doctor and read the network footprint. Every mgimind port is loopback-bound, and mgimind changes no route, DNS or firewall setting, so it cannot pull traffic out of a tunnel. Look at what starts alongside it: Docker Desktop and WSL2 add virtual adapters of their own, and TUN clients with strict routing can drop the tunnel when an adapter appears or disappears.

Security

• Downloads verify against pinned SHA-256 (fail-closed) for ONNX Runtime (linux-x64), Qdrant, and the default models (e5 and the reranker). Other platforms and custom models warn instead of trusting blindly.
• Qdrant binds to 127.0.0.1 only and supports an API key.
• The vault is AES-256-GCM with an Argon2id-derived key (parameters pinned so a library upgrade can't lock you out). Terminal-only, so the master password and decrypted secrets never travel over the MCP channel. The mind_vault_* MCP tools return terminal instructions, never the secret value.
• File writes are atomic and directory-fsynced, so a crash leaves the old file or the new one, never a corrupt one.

Network footprint

Every socket mgi-mind opens is bound to loopback:

serve-http --host is the one way to bind another interface (a Docker -p mapping needs it), and a non-loopback host is refused unless you also pass --agent-token.

mgimind installs no driver, adds no network adapter, and changes no route, DNS or firewall setting. mgimind doctor prints this footprint with a live probe of every port, so the claim is something you can check rather than trust.

Status and audit

Current version: 2.8.2 (semver-stable since v1.0.0). The 0.x line built the foundation: the audit log and ephemeral viewer (0.10), the quarantine layer and best-effort retrieval policy (0.11), the viewer wave (0.12), session liveness (0.13), and procedural memory (0.14, benchmarked on LongMemEval-S plus a 227-pair error→fix dataset from 20 public repos). The 1.x line added the validity model: the duel rule, the doubt window and its background re-test, bi-temporal fact supersession, the confidence score, and typed outcome signals (mind_outcome), plus install-mode CPU/GPU profiles.

The MCP surface is additive-only until a 2.0 bump: 29 live tools plus 15 deprecated aliases kept for compatibility. Nothing is renamed or removed inside 2.x; new verbs (most recently mind_skill) may be added. The other 1.0 contracts are the asymmetric "Qdrant now → md says" reconcile diff and the MGIMIND_MODEL_VARIANT={cpu|gpu|auto} switch.

The project went through a code audit; AUDIT_STATUS.md accounts for every issue one by one with where it was fixed or why it was deferred. CHANGELOG.md has the per-release history for the 1.x line plus the v1.7 candidate section (pre-1.0 releases are in CHANGELOG-ARCHIVE.md), and ROADMAP.md names what is committed for the next releases and which directions are still candidate at the v3.0 horizon.

Project layout

src/  cli.rs         command dispatch and output rendering  storage.rs     Qdrant: single collection, hybrid search, history, stats, migrate, chunking  embedder.rs    ONNX embedding (model-aware pooling, prefixes, 512-token cap)  reranker.rs    cross-encoder reranking  knowledge.rs   knowledge-graph facts + cardinality + supersession  duel.rs        duel rule: resolve contradicting facts (flip / contested / quarantine)  doubt.rs       doubt window + background active re-test of entrenched facts  activity.rs    recent-query buffer, corpus mean (centering), origin-context store  confidence.rs  per-fact confidence score (dependants / confirmations / signals)  outcome.rs     typed external signals (test_passed, code_compiled, ...) into weight  procedure.rs   procedural memory: learn / recall / outcome  ingest.rs      auto-extract & ingest candidates  relevance.rs   relevance gate (length, blacklists, decision markers, token novelty)  retrieval_policy.rs  search-before-answer classifier  provenance.rs  cited external snippets with mandatory source  extractor.rs   optional local LLM extraction (off by default)  consolidate.rs merge duplicates, report cold entries  md_reconcile.rs md import as reconcile with "md wins"  audit.rs       append-only audit log for memory/fact/procedure/skill                 writes, updates, deletes and quarantine transitions (not                 pinned blocks, kv, vault, or sessions yet)  http_api.rs    loopback HTTP surface for multi-agent access  viewer.rs      ephemeral local HTTP viewer (axum, static frontend baked in)  pulse.rs       live graph pulses for the viewer  session.rs     per-agent session files  secrets.rs     secret-scrub on the write path  vault.rs       encrypted secret vault (terminal only)  mcp.rs         MCP server over stdio (hand-rolled JSON-RPC; warm in-process models)  config.rs      configuration + install-mode profiles  integrity.rs   pinned SHA-256 hashes for downloads  util.rs        atomic writes, verified downloads  (plus migrate, install-mode, access/decay, and bench harnesses)tests/  cli_integration.rs   black-box tests against a real Qdrant (CLI + MCP round-trip)  http_integration.rs  the multi-agent HTTP surface end to end

License

Apache-2.0. See LICENSE.