Semantic memory for AI agents with hybrid search, knowledge graph, and consolidation
Semantic memory for AI agents with hybrid search, knowledge graph, and consolidation
neuromcp · v0.4.2
by AdelElo13
neuromcp — Sovereign Memory for AI agents
Any model. Your memory. Stays local.
neuromcp is the first Sovereign Memory layer for AI: an open-source MCP server that gives Claude, GPT, Gemini, and Ollama persistent, searchable memory — stored entirely on your machine. No API keys. No cloud sync. No subscription required to remember who you are.
Sovereign Memory = data that you own outright, lives on hardware you control, and is portable across every model you use. Cloud memory products own your data; Sovereign Memory means you do.
npx neuromcp
Why neuromcp
The LLM is a commodity. Your memory is the moat.
GPT-5, Claude 4, Gemini — they all converge. The model you use next year will differ. The memory of every conversation, decision, and preference you build is yours. neuromcp keeps that layer on your machine and makes it portable across any MCP-compatible client.
Local-first is a design choice, not a limitation.
No telemetry. No data leaves your laptop. No vendor has a copy of your conversations. Audit every line of code that touches your memory. SQLite + local embeddings; everything fits on one disk.
One install. Every client.
Claude Desktop, Cursor, Windsurf, Codex CLI, Continue, LibreChat, Open WebUI — neuromcp speaks MCP, so it works wherever MCP is supported. Switch models tomorrow; your memory follows.
Real recall, not keyword matching.
Hybrid retrieval combines vector search (nomic-embed-text, 768-dim), BM25 full-text, graph links, and a learned usefulness prior. At 500 distractors on LongMemEval, R@5 holds at 93.3%. Your context window gets the right memory, not just the most recent.
LongMemEval-S accuracy
| Run | Score | Sample | Config |
|---|---|---|---|
| v7 (current) | 96.08% (98/102) | n=102 | Opus generator + Opus judge, single-model |
| v6 | 95.10% (97/102) | n=102 | Same as v7, prior hint set |
Repro: OMB_ANSWER_LLM=claude OMB_ANSWER_MODEL=opus OMB_JUDGE_LLM=claude OMB_JUDGE_MODEL=opus uv run omb run --dataset longmemeval -s s -m neuromcp -c "single-session-user,single-session-assistant,multi-session,temporal-reasoning,knowledge-update,single-session-preference" --query-limit 17
Sample size honesty. n=102 (17 per category × 6 categories). Wilson 95% CI for 98/102 ≈ 90.5–98.7%. Full 500q run with the same config is the next milestone before any "top-tier" claim.
Benchmarks (v0.18.0)
Oracle split (clean — easy mode)
| Mode | R@5 | R@10 | Hit Rate |
|---|---|---|---|
| Extracted (hybrid) | 100% | 100% | 100% |
Oracle-split LongMemEval isolates the correct memory in a small
corpus. Every local MCP memory system claims ~99% here. It measures
"does the ranker work on clean inputs" — nothing more.
Distractor split (v0.18.0, honest)
Same 30 questions + 1000 random distractor memories drawn from other
questions' haystacks. The correct memory now competes against real noise.
| Embedder | Distractors | N | R@5 | R@10 | MRR |
|---|---|---|---|---|---|
Ollama nomic-embed-text |
0 (oracle) | 30 | 100% | 100% | 100% |
Ollama nomic-embed-text |
200 | 5 | 100% | 100% | 100% |
Ollama nomic-embed-text |
500 | 30 | 93.3% | 93.3% | 80.3% |
Ollama nomic-embed-text |
1000 | 5 | 100% | 100% | 74% |
Reproduce: npx tsx eval/longmemeval-distractor-runner.ts --limit 5 --distractors 1000
Sample sizes. The 500-distractor row is n=30 (Wilson 95% CI for
28/30 ≈ 78-99% R@5). The 1000-distractor row is n=5 — preliminary,
Wilson 95% CI [57%, 100%]. The 1000-distractor n=30 run takes ~36 min
on a single Ollama instance; cached-distractor batching is v0.19.0
work. Treat 500-distractor numbers as defensible, 1000-distractor as
directionally positive but underpowered.
Head-to-head comparison is explicit v0.19.0 work. Hindsight (local
OSS MCP, ~94.6% LongMemEval claimed) and Mem0/Zep publish their own
numbers on their own harnesses. Until we run all of them against the
same corpus + embedder, calling any local MCP server "state of the art"
is marketing, not measurement. neuromcp publishes its numbers with
sample-size caveats so you can judge direction; don't read absolute
superiority into them yet.
Hybrid ranker (BM25 + vector + attention + graph + usefulness prior)
keeps R@5 = 100% at 1000:1 distractor:target ratio on the observed
sample. MRR drops to 74% because the correct memory is sometimes not
rank-1 but always rank ≤ 5 in what we saw. Earlier v0.18.0 numbers
(R@5 23%) were from a test FakeEmbedder — fixed in v0.18.1.
What this benchmark does NOT prove: end-to-end answer
correctness, long-horizon multi-session reasoning, or superiority
over commercial cloud systems (Mem0, Zep) on their own benchmarks.
Those comparisons need their numbers on the same distractor split,
which hasn't been published.
Why
AI agents forget everything between sessions. Existing solutions either store flat key-value pairs (useless for real knowledge) or require cloud infrastructure and API keys.
neuromcp gives you two layers of memory:
- MCP Server — hybrid search (vector + full-text + graph), verbatim recall, memory governance, automatic consolidation, all in a single SQLite file
- Wiki Knowledge Base — compiled Markdown knowledge that survives crashes, compounds over sessions, and gives your agent project-aware context at every startup
Inspired by Karpathy's LLM Wiki, Mastra's Observational Memory, and Zep's temporal knowledge graphs — but simpler than all of them. No vector DB, no embeddings pipeline, no cloud. Just Markdown files + Git + hooks.
Architecture
~/.neuromcp/
├── memory.db ← SQLite: hybrid search, MCP tools
├── wiki/ ← Compiled knowledge (git-tracked)
│ ├── index.md ← Routekaart — LLM reads this FIRST
│ ├── schema.md ← Operating rules for the LLM
│ ├── log.md ← Append-only changelog
│ ├── people/ ← User profiles, preferences
│ ├── projects/ ← Project knowledge (stack, auth, URLs)
│ ├── systems/ ← Infrastructure (tools, MCP servers)
│ ├── patterns/ ← Reusable patterns (error fixes, routing)
│ ├── decisions/ ← Architecture decisions with context
│ └── skills/ ← Repeatable procedures
└── raw/sessions/ ← Raw session logs (auto-generated)
How the wiki works
| When | What happens |
|---|---|
| Session start | Hook injects index.md + user profile + auto-detected project page (~1300 tokens) |
| During session | LLM updates wiki pages when learning something persistent |
| Every 8 tool calls | Hook reminds LLM to update the wiki |
| Session end | Hook writes raw session log + git auto-commits all wiki changes |
| Crash | Checkpoint every 5 tool calls to file. Git history for rollback. |
Self-healing consolidation pipeline (v0.15.0+)
Every ~4h the launchd agent runs run-consolidation.sh, which
orchestrates four steps end-to-end:
consolidate-sessions.py— batches raw sessions per project,
asks Claude for a factual summary, and fact-checks it against the
raw sources. When the auditor flags specific unsupported claims the
consolidator now auto-strips those lines and re-audits once — so
one speculative sentence no longer kills a whole batch.rescue-rejected.py— any batch that still fails is parsed,
the unsupported claims are removed, and the cleaned summary is
appended to its wiki page. Pure text surgery, no LLM calls.entity-linker.py— cross-links every page: a bare-word mention
of another registered entity (people/, projects/, systems/) is added
to the page'srelated:frontmatter. Makes the wiki act like a
graph without a separate graph database.rebuild-index.py— regeneratesindex.mdand per-category
-index.mdfiles. Categories over 10 pages are auto-split so the
session-start router stays compact as the wiki scales.
The pipeline is idempotent — safe to re-run at any time.
What the LLM knows at session start
Schema (operating rules) → How to maintain the wiki
Index (knowledge map) → What knowledge exists
User profile → Who you are, how you work
Project page → Current project details (auto-detected from cwd)
Last session → What happened last time
Quick Start
1. Start the MCP server
npx neuromcp
2. Initialize the wiki + hooks (required for closed-loop attribution)
npx neuromcp-init-wiki
This creates the wiki structure, installs hooks (Claude Code) and rules (other editors), and configures everything automatically. Without this step, npx neuromcp still runs as a plain MCP server with 42 tools, but the critic hook that closes the attribution loop is not installed — retrieval works but usefulness scores never accumulate. Safe to run multiple times — won't overwrite existing config.
Editor Compatibility
neuromcp works with any MCP-compatible editor. Two tiers of integration:
| Feature | Claude Code | Cursor / Windsurf / Cline / Copilot / JetBrains / Zed |
|---|---|---|
| MCP tools (40+) | Full | Full |
| Context at session start | Hooks (automatic) | Rules (LLM-driven, best-effort) |
| Persist at session end | Hooks (automatic) | Rules (LLM-driven, best-effort) |
| Wiki reminders | Every 8 tool calls | No |
| Crash-resilient checkpoints | Yes | No |
Claude Code gets the full experience via native hooks — context injection and persistence happen automatically, even if the LLM forgets.
Other editors get rules files that instruct the LLM to call neuromcp tools at session start/end. This depends on LLM compliance — it works well in practice but is not guaranteed like hooks.
# Auto-detect installed editors
npx neuromcp-init-wiki
# Target a specific editor
npx neuromcp-init-wiki --editor cursor
# Install rules for all supported editors
npx neuromcp-init-wiki --editor all
Supported editors: cursor, windsurf, cline, copilot (VS Code), jetbrains, zed
Recommended: Add Ollama for real semantic search
ollama pull nomic-embed-text
neuromcp auto-detects it. No config needed.
Installation
Claude Code
// ~/.claude.json → mcpServers
{
"neuromcp": {
"type": "stdio",
"command": "npx",
"args": ["-y", "neuromcp"]
}
}
Claude Desktop
// ~/Library/Application Support/Claude/claude_desktop_config.json
{
"mcpServers": {
"neuromcp": {
"command": "npx",
"args": ["-y", "neuromcp"]
}
}
}
Cursor / Windsurf / Cline
Same format — add to your editor's MCP settings.
Per-project isolation
// .mcp.json in project root
{
"mcpServers": {
"neuromcp": {
"type": "stdio",
"command": "npx",
"args": ["-y", "neuromcp"],
"env": {
"NEUROMCP_DB_PATH": ".neuromcp/memory.db",
"NEUROMCP_NAMESPACE": "my-project"
}
}
}
}
MCP Surface
Core Tools
| Tool | Description |
|---|---|
store_memory |
Store with semantic dedup, contradiction detection, surprise scoring, entity extraction. |
search_memory |
Hybrid vector + FTS search with RRF ranking, graph boost, cognitive priming. Returns explain metadata (trust, contradictions, claims, confidence). |
recall_memory |
Retrieve by ID, namespace, category, or tags — no semantic search. |
forget_memory |
Soft-delete (tombstone). Supports dry_run. |
consolidate |
Dedup, decay, prune, sweep. commit=false for preview, true to apply. |
memory_stats |
Counts, categories, trust distribution, DB size. |
export_memories |
Export as JSONL or JSON. |
import_memories |
Import with content-hash dedup. |
search_all |
Unified search across extracted memories and verbatim text with source labels. |
Verbatim Tools
| Tool | Description |
|---|---|
store_verbatim |
Store raw conversation text — no summarization, never pruned. |
search_verbatim |
Full-text search (FTS5) on verbatim entries for exact recall. |
verbatim_stats |
Stats on verbatim storage: total entries, size, distribution. |
Resources (13)
| URI | Description |
|---|---|
memory://stats |
Global statistics |
memory://recent |
Last 20 memories |
memory://namespaces |
All namespaces with counts |
memory://health |
Server health + metrics |
memory://stats/{namespace} |
Per-namespace stats |
memory://recent/{namespace} |
Recent in namespace |
memory://id/{id} |
Single memory by ID |
memory://tag/{tag} |
Memories by tag |
memory://namespace/{ns} |
All in namespace |
memory://consolidation/log |
Recent consolidation entries |
memory://operations |
Active/recent operations |
Prompts (3)
| Prompt | Description |
|---|---|
memory_context_for_task |
Search relevant memories and format as LLM context |
review_memory_candidate |
Show proposed memory alongside near-duplicates |
consolidation_dry_run |
Preview consolidation without applying |
Wiki Knowledge Base
The wiki is the compiled, human-readable knowledge layer. It replaces the chaos of session logs with structured, interlinked Markdown pages.
Why a wiki instead of more vector search?
| Traditional RAG | neuromcp Wiki |
|---|---|
| Re-derives answers every query | Knowledge compiled once, refined over time |
| Chunking artifacts, retrieval noise | Human-readable pages with source citations |
| Vector DB, embedding pipeline | Plain Markdown + Git |
| Black box retrieval | Auditable, editable, portable |
| Knowledge evaporates | Knowledge compounds |
Wiki page format
---
title: My Project
type: project
created: 2026-04-06
updated: 2026-04-06
confidence: high
related: [other-project, oauth-setup]
---
# My Project
Description, stack, auth, deployment details...
How to use
The wiki works automatically once hooks are installed. The LLM:
- Reads
index.mdat session start to know what knowledge exists - Reads specific pages when relevant to the current task
- Updates pages when learning something new
- Gets reminded every 8 tool calls if the wiki needs updating
You can also browse and edit the wiki manually — it's just Markdown files.
Auto-consolidation (optional)
Once you accumulate raw session logs, the wiki can be kept fresh automatically. A scheduled job reads unprocessed sessions, groups them per project (by detecting $HOME/projects/<name> paths in the session content), and uses the claude CLI to synthesise a ## [date] entry into the right wiki page.
npx neuromcp-enable-consolidation
What it installs:
~/.neuromcp/scripts/consolidate-sessions.py— the worker~/.neuromcp/scripts/run-consolidation.sh— threshold-guarded runner- macOS: a launchd agent that fires every 4 hours (
com.neuromcp.consolidate) - Linux: prints a cron snippet to add manually
Requirements:
python3≥ 3.8 onPATH- the
claudeCLI onPATH
Guards built in:
- Threshold: skip if fewer than 5 unprocessed sessions
- Output is extracted from a fenced markdown block; apology/narration text is rejected
- Ledger (
~/.neuromcp/consolidation-ledger.json) makes re-runs idempotent - Large project backlogs are auto-batched (default 15 sessions per
claudecall; override with--max-sessions)
Uninstall: npx neuromcp-enable-consolidation --uninstall
Change interval: npx neuromcp-enable-consolidation --interval 7200 (every 2 hours)
Hallucination guard (eval-loop). Every consolidator output goes through a second Haiku audit before the wiki is touched. If any factual claim in the generated summary is not traceable to the raw sessions, the chunk goes to ~/.neuromcp/review-queue/ instead of the wiki. No hallucinated claims leak through.
Atomic facts with temporal supersession. After a summary is approved, it is also distilled into short standalone facts and stored as category='fact' rows with valid_from=today. When a new fact is Jaccard-similar to an existing one in the same project, Haiku decides whether NEW supersedes OLD — if yes, the old row gets superseded_by_id and valid_to set. Retrieval defaults to current facts only (superseded_by_id IS NULL), so outdated conclusions never resurface.
Auto-retrieve + hybrid indexing
Once the wiki has content, make it searchable so the UserPromptSubmit hook can surface relevant pages automatically (no more "LLM must remember to call search"):
npx neuromcp-index-wiki # index wiki pages into memories_fts + memories_vec
npx neuromcp-index-wiki --rebuild # wipe wiki entries first, then reindex
npx neuromcp-index-wiki --dry-run # preview what would change
npx neuromcp-index-wiki --no-embed # FTS-only mode (no embedding provider needed)
npx neuromcp-backfill-embeddings # embed any memory still missing a vector
The indexer splits each page on ## section headers and stores every section as a deduplicated memory (source='wiki', category='wiki'). Each section is both written to the FTS5 index and embedded via the configured provider (Ollama → OpenAI → ONNX) so vector search works too.
At prompt time the neuromcp-auto-retrieve.js hook calls neuromcp-query, which runs FTS5 BM25 and sqlite-vec cosine search in parallel and fuses the rankings via Reciprocal Rank Fusion (k=60). The top-3 merged results are injected as <neuromcp-recall> context.
The hook is installed automatically by neuromcp-init-wiki and registered under UserPromptSubmit in Claude Code's settings.json. Re-run the indexer after large wiki updates (or schedule it — it's idempotent).
Tuning:
| Env var | Default | Purpose |
|---|---|---|
NEUROMCP_BM25_THRESHOLD |
-1.0 |
Stricter (more negative) = fewer weak keyword matches |
NEUROMCP_QUERY_BIN |
auto-detect | Override the neuromcp-query binary path |
NEUROMCP_NO_EMBED |
0 |
Set to 1 to force FTS-only indexing |
NEUROMCP_CONTRADICTION_CHECK |
1 |
Set to 0 to skip Haiku supersession judgments |
NEUROMCP_AUDIT_FAIL_OPEN |
0 |
Set to 1 to bypass the consolidator audit on infrastructure failure (default is fail-CLOSED) |
Known upstream issues
memories_vec does not reclaim space after DELETE — sqlite-vec #54 / #265. When you re-index after editing wiki sections, the old vector rows are marked deleted but their storage stays. The database file grows monotonically until you run npx neuromcp-index-wiki --rebuild, which drops and re-creates the vector rows. Run a rebuild every few weeks if you edit the wiki heavily.
claude CLI streaming hangs from non-TTY subprocesses on macOS — if you script interactions with claude -p from another process (e.g. scheduled jobs), pipe it through script -q /dev/null to allocate a pseudo-TTY. Without that the stdout buffer never flushes. We work around this inside the consolidator where needed.
Memory Governance
Namespaces isolate memories by project, agent, or domain.
Trust levels (high, medium, low, unverified) rank search results and control decay resistance.
Soft delete tombstones memories — recoverable for 30 days.
Content hashing (SHA-256) deduplicates at write time.
Lineage tracking records source, project ID, and agent ID per memory.
Configuration
All via environment variables. Defaults work for most setups.
| Variable | Default | Description |
|---|---|---|
NEUROMCP_DB_PATH |
~/.neuromcp/memory.db |
Database file path |
NEUROMCP_EMBEDDING_PROVIDER |
auto |
auto, onnx, ollama, openai |
NEUROMCP_DEFAULT_NAMESPACE |
default |
Default namespace |
NEUROMCP_AUTO_CONSOLIDATE |
false |
Enable periodic consolidation |
NEUROMCP_TOMBSTONE_TTL_DAYS |
30 |
Days before permanent sweep |
NEUROMCP_LOG_LEVEL |
info |
debug, info, warn, error |
What's New in v0.9
Auto-Capture (v0.9.0)
Session hooks automatically extract high-signal events — no manual store_memory calls needed:
| Detected | Category | How |
|---|---|---|
| CronCreate / ScheduleWakeup calls | intent |
Regex on transcript |
| "Remember this" / "Onthoud dit" | decision |
Pattern matching |
| Domain monitoring (whois checks) | intent |
Command detection |
| Key decisions ("we decided...") | decision |
Language patterns |
| Deployments (npm publish, etc.) | event |
Command detection |
Full Pipeline Auto-Capture (v0.9.1)
Auto-captured memories now go through the full store pipeline: dedup, contradiction detection, embeddings, entity extraction, and claims — via HTTP endpoint (POST /api/store). Falls back to raw SQL when HTTP is unavailable.
Contradiction resolution now has three tiers:
- Supersede (score > 0.5): old memory invalidated, new one takes over
- Coexist (score 0.35–0.5): both kept, linked via
contradictsedge in knowledge graph - Flag (score 0.3–0.35): reported for review
Explain Mode (v0.9.2)
Every search_memory result includes an explain field:
{
"explain": {
"source_trust": { "level": "high", "reason": "Directly provided by user" },
"temporal_validity": { "currently_valid": true, "superseded_by": null },
"contradictions": [{ "memory_id": "abc", "content_preview": "...", "resolution": "coexist" }],
"claims": [{ "subject": "neuromcp", "predicate": "version", "object": "0.9.2" }],
"confidence": { "retrieval_score": 0.016, "source_trust_score": 1.0, "overall": 0.85 }
}
}
We publish all of this — schema versions, consolidation math, critic output, benchmark numbers with CIs — so you can audit exactly what the system remembers and how. If another local-first system publishes the same or better, link welcome.
Comparison
| Feature | neuromcp | Hindsight | Mem0 | Letta/MemGPT | agentmemory |
|---|---|---|---|---|---|
| LongMemEval R@5 (oracle) | 99.8% | — | — | — | — |
| LongMemEval R@5 (1000 distractors, n=5, Ollama) | 100% (preliminary, CI [57%, 100%]) | not published | not published | not published | not published |
| Search | Hybrid (vector + FTS + RRF + graph) | Vector + rerank | Vector | Vector | Vector |
| Auto-capture | Deterministic (no LLM cost) | LLM extraction | No | Agent self-edit | Yes |
| Explain mode | Yes (trust, contradictions, claims) | No | No | No | No |
| Knowledge graph | Entities, relations, PageRank | Entities + beliefs | No | No | No |
| Contradiction detection | 3-tier (supersede/coexist/flag) + graph edges | Belief updating | No | No | No |
| Temporal validity | valid_from/valid_to on memories + relations | Yes | No | No | No |
| Wiki knowledge base | Compiled Markdown + Git | No | No | Tiered blocks | No |
| Local-first | SQLite, zero cloud | SQLite | Cloud / Postgres | Server | Local |
| Embeddings | Built-in ONNX (zero config) + Ollama | External | External API | External | External |
| Governance | Namespaces, trust levels, soft delete | Namespaces | API keys | Agent-scoped | Cross-agent |
| Infrastructure | Zero | Zero | Cloud account | Server | Zero |
| Pricing | Free (AGPL-3.0) | Free (MIT) | Freemium ($23.9M funded) | Free ($10M funded) | Free (Apache-2.0) |
License
AGPL-3.0 for the engine in src/. MIT for bin/, templates/,
scripts/, docs/, and examples/ (carve-out — see LICENSE-EXAMPLES).
License FAQ
Can I use neuromcp commercially? Yes. Running neuromcp as part of your
own application, on your own infrastructure, is unrestricted. AGPL only
imposes obligations if you modify the engine code AND distribute or
host it as a network service.
Can I install neuromcp from npm in my closed-source product? Yes. Using
the published binary as a dependency does not trigger AGPL contagion.
What if I host neuromcp as a SaaS? Then AGPL §13 applies: you must make
the source code (including your modifications) available to your users.
This is the explicit anti-fork clause we chose for the engine — it stops
well-funded competitors from taking the code, putting it behind a login,
and shipping it as their own product.
Can I copy a CLI script or template? Yes. Everything in bin/,
templates/, scripts/, docs/, and examples/ is dual-licensed
AGPL-3.0 OR MIT. Pick MIT in your downstream project.
Need different terms for the engine? Commercial dual-license is
available — contact the maintainer.