NEWOne memory for every agent, bot and chat windowOne memory for every agent and botWhat’s new →

Demo

See it
in action.

Rent the LLM. Own the memory. Watch the full eight-minute demo, then see the real dashboard: an answer check that says “I don’t have that”, Recall Lab, memory kinds, saved views and Connected apps.

Watch the Demo

One memory for every AI agent and bot. The full demo.

Claude Code, Codex, Cursor, ChatGPT and Grok Bot share one local memory. Every tab, light and dark: recall, the knowledge graph, the answer check, SLM Mesh, governance, caching and compression.

Open on YouTube →

The dashboard

Real screens.
Nothing mocked.

Captured from a real install in Mode A, with Laya running on the Mac and a store of fictional project memories. Each verdict, score and timing is what SLM returned.

Memories that answer the question: Laya judges that the memories answer “When is Project Kestrel going live?” with confidence 0.85, in 1.4 s of the 3 s limit. — **Memories that answer the question**Laya judges that the memories answer “When is Project Kestrel going live?” with confidence 0.85, in 1.4 s of the 3 s limit.

Memories that do not: “I don’t have that”: For “How much did the Kestrel pilot cost?” Laya finds no memory that answers it, and SLM says it doesn’t have that instead of guessing. — **Memories that do not: “I don’t have that”**For “How much did the Kestrel pilot cost?” Laya finds no memory that answers it, and SLM says it doesn’t have that instead of guessing.

Recall Lab: why each memory was chosen: Per-channel scores for “what did we decide about the database”. — **Recall Lab: why each memory was chosen**Per-channel scores for “what did we decide about the database”.

Every memory, with its kind and project: The Memories table filtered by kind: decisions, rules, corrections and facts, with the project each belongs to. — **Every memory, with its kind and project**The Memories table filtered by kind: decisions, rules, corrections and facts, with the project each belongs to.

Saved views: A saved question that runs the same search your agent uses and shows the memory each result came from. — **Saved views**A saved question that runs the same search your agent uses and shows the memory each result came from.

Connected appsWeb access is on, linked with GitHub for the default profile, renews automatically, with a free daily allowance and a button to add an app or turn access off.

Interactive Demo

Pick a capability.

Three documented command flows. Click a tab to inspect the interaction, then validate the same command on your installed release.

superlocalmemory — interactive demo

Use Cases

Built for how you actually work.

Whether you're on a team, solo, or building agents — SLM slots into your workflow without friction.

01 · Engineering Teams

Engineering Teams

Persist architecture decisions, API contracts, and bug resolutions across sessions, then inspect what the next session actually recalls.

$ slm remember "API rate limit: 100 req/min per user, burst 200" ✓ Queryable → operation <opaque-id> · enrichment pending $ slm recall "API limits" ↓ rate: 100/min, burst: 200 · relevance and stored confidence reported separately

02 · Solo Developers

Solo Developers

Build a searchable memory of everything you've ever debugged, decided, or documented. Your personal engineering knowledge base.

$ slm recall "how did we fix the CORS issue last month?" ↓ Added Origin header allowlist in nginx.conf · inspect relevance and stored confidence fields

03 · AI Agent Pipelines

AI Agent Pipelines

Give autonomous agents persistent context that survives session boundaries through a local core memory path; optional providers and connectors are configured separately.

$ slm recall "current sprint goal" --json {"data": {"fact": "ship auth refactor by June 30", ...}}

Benchmark Results

Measured. Published. Reproducible.

Published LoCoMo evidence is carried over with its original evaluation scope disclosed beside every number.

74.8%

Mode A Retrieval

10 CONVERSATIONS · 1,276 QUESTIONS · GPT-4.1-MINI ANSWER SYNTHESIS

87.7%

Mode C cloud-assisted

81 QUESTIONS · CONV-30 · CLOUD EMBEDDINGS + GPT-4.1-MINI

Content-specific

Compression behavior

SAFE CODE/JSON PATHS MAY YIELD NO REDUCTION

Cost per cache hit

EXACT-MATCH · SESSION TTL

LoCoMo is a long-context conversation memory benchmark. Results published in arXiv:2603.14588.

Quick-Start Configs

Explicit client configuration.

Three lines or fewer to connect SLM to your tool of choice. Copy, paste, done.

// .claude/settings.json "mcpServers": "superlocalmemory": "command": "slm", "args": ["mcp"]

// ~/.cursor/mcp.json "mcpServers": "superlocalmemory": "command": "slm", "args": ["mcp"]

# Add to ~/.zshrc or ~/.bashrc alias remember='slm remember' alias recall='slm recall' # Then just: remember "Alice owns the auth service" recall "who owns auth?"

Ready to try it?

Two commands: npm install -g superlocalmemory, then slm setup. The core is free, open source and local-first.

Install SuperLocalMemory Read the Docs

See itin action.