Everything in one page · mnueron docs

The all-in-one preview of mnueron's documentation. Every capability is on this page so you can scan the whole product before deciding what to read deeply.

What mnueron is

One memory layer. Every AI tool. Every app you build. mnueron gives every AI you use the same long-term memory — Claude Desktop, Claude Code, Cursor, Windsurf, Cline, claude.ai, chatgpt.com, and any custom app you build with an OpenAI / Anthropic / Mistral / Gemini key.

Modes:

Local mode (free, forever): memories live at ~/.mnueron/memories.db on your machine. No account, no telemetry. Install the npm CLI and the setup wizard wires it into every AI tool you have installed.
Hosted mode (paid): memories sync across machines via https://www.mnueron.com. Free signup; paid tiers for higher volume + premium LLM features (entities, relations, runbook proposals, consolidation).

5-minute install

node --version    # need 20+
npm install -g mnueron
mnueron setup     # interactive wizard, detects + wires every AI tool

Restart whichever AI tool was wired. Open a new chat and ask: "what MCP tools do you have available?" — you should see memory_save, memory_recall, memory_list, memory_get, memory_delete, memory_get_thread, plus the procedural-memory tools.

Hosted-mode quick start

Sign up at https://www.mnueron.com.
Visit /account-settings/tokens → create a token → copy the mnu_... value.
Use it as a bearer header against /api/... endpoints.

curl -H "Authorization: Bearer mnu_..." \
     "https://www.mnueron.com/api/memories?limit=5"

The six memory types

mnueron stores six conceptually distinct kinds of memory. All six live in the same memories table — what differs is the metadata and the recall path.

Type	Captures	Recall pattern
Episodic	Raw conversations as they happened	`memory_get_thread(parent_ref)` rebuilds a conversation; `memory_recall` over the chunked turns
Semantic	Standalone facts	BM25 + (optional) vector search across `memories`
Entity	Canonical people, projects, places	`/api/entities` resolves "M. Mangal" ≡ "Maya Mangal" ≡ "Maya"
Relational	Who knows whom, what depends on what	`/api/relations` returns the graph slice — viewable on `/dashboard/graph`
Procedural	"Here's how to do X" runbooks	`/api/recall/unified` matches trigger phrases; chat agent emits ```runbook``` JSON cards
Consolidation	"This is a duplicate" proposals	`/api/consolidate?action=propose` finds overlaps; `/dashboard/consolidate` reviews

Read the memory types explainer for the deep version with examples.

REST API

Verb	Path	Purpose
POST	`/api/memories`	Save. Auto-synopsis fires when content > 800 chars and gate is on.
GET	`/api/memories?q=...`	Full-text search with date + metadata filters.
GET	`/api/memories/:id`	Single memory by id. Auto-logs a recall.
PATCH	`/api/memories/:id`	Partial update. Metadata merged. History audit trail.
DELETE	`/api/memories/:id`	Remove (RLS-scoped).
POST	`/api/memories/search/bulk`	Up to 25 queries in one round-trip.
GET	`/api/namespaces`	List your namespaces with counts.
POST	`/api/recall/unified`	Hybrid: returns matching memories AND runbooks.
POST	`/api/procedural`	Save a runbook.
GET	`/api/procedural`	List runbooks.
GET	`/api/entities`	List + filter entities (paid).
GET	`/api/relations`	Graph slice (paid).
GET, POST	`/api/consolidate`	Duplicate-merge queue + apply (paid).
POST	`/api/chat`	One-shot dashboard chat completion.
GET, POST	`/api/chat/sessions`	Chat session CRUD.
GET, POST, PUT, DELETE	`/api/webhooks`	HMAC-SHA256 signed deliveries.
GET	`/api/health`	Liveness probe (no auth).
POST, GET, DELETE	`/api/auth/tokens`	Token CRUD.

All write endpoints honor your org's quota; LLM-backed endpoints (chat, entity extraction, consolidator merge) require either a paid plan or a BYOK key (x-anthropic-key / x-openai-key headers).

SDKs

Python — pip install mnueron
TypeScript — npm install @mnueron/sdk
.NET / C# — drop MnueronClient.cs into any project
VS Code — install mnueron — AI memory from the Marketplace

Identical surface across all four. See the SDK overview.

Integrations (auto-wired by `mnueron setup`)

Claude Desktop, Claude Code, Cursor, Windsurf, Cline
Planned: Continue.dev, Zed, Aider, Goose, OpenCode
Chrome / Firefox extension (claude.ai, chatgpt.com, gemini.google.com)
VS Code extension panel

Dashboard surfaces (mnueron.com)

/dashboard — memory list, filters, pagination, decay + recall chips
/dashboard/chat — conversational chat agent backed by your memories
/dashboard/analytics — KPIs, time-series, leaderboard
/dashboard/entities — entity canonical list (paid)
/dashboard/entities/[id] — entity detail with linked memories
/dashboard/graph — force-directed relation graph (paid)
/dashboard/procedural — runbook library
/dashboard/consolidate — duplicate-review queue (paid)
/dashboard/context-builder — assemble a context bundle from saved memories

Pricing

Free Local — $0 forever, fully featured local mode.
Personal — $9/mo, hosted sync + cross-machine.
Pro — $19/mo, unlimited memories, retention, audit log.
Team — $25/user/mo (min 3), shared namespaces, RBAC.
Enterprise — custom, SSO, BAA, SLA.