AgentMemory

Give your agent long-term memory

Long-term memory that survives across agent runs.

Deprecated — frozen runtime

This primitive still works for the operators already on it, and that is a compatibility promise rather than a recommendation. Do not build new work on it. Use instead: the **brain**: `naive brain remember` / `naive brain attach`, or `self.brain` in-agent.

Get your API keyRead the docs
deep researchlive
$search "market size…"
search
extract
read
cross-ref
synthesize
cite

Paste and ship

Drop this into Cursor, Claude Code, or any agent. It reads the skill manifest and wires up Memory for you.

Agent prompt

Read https://usenaive.ai/skill.md and use the Naïve Memory primitive in my project. Start with: naive memory add

What you get

Long-term storeFacts persist across sessions.
Semantic retrievalFetch relevant context.
Per-user scopeTenant-isolated memory.
CEO integrationCompany-wide knowledge base.

One call to AgentMemory

CLI, SDK, or REST — one bearer token, one credit balance.

CLI

Memory

Store and retrieve long-term memory for agents — facts, preferences, and state that survives across runs.

Long-term store
Facts persist across sessions.
Semantic retrieval
Fetch relevant context.
Per-user scope
Tenant-isolated memory.
CEO integration
Company-wide knowledge base.
# Run with your API key exported
$ naive memory add

Start building with AgentMemory

Get your API keyRead the docs →

One API key. One credit balance. Every primitive.

Related capabilities

Agent Brain300+ models, live research, a citable knowledge base, and lasting memory.View →All primitives in this batchLLM, search, brain, and memory.View →Lab: MemoryHow durable memory should work for agents that outlive a single run.View →

Related primitives

AgentLLM

Chat completions across 300+ models through one router.

AgentStaff

Deprecated

Hire and configure AI workers.

Frozen. Use instead: declare a role with `agent({…})` inside `team({ lead, agents })`, then `naive up`

AgentCEO

Deprecated

Turns company goals into delegated agent work.

Frozen. Use instead: a `team({ lead })` on `runtime.durable()`; work is submitted with `naive teams submit`