The agent harness performance optimization system. Skills, instincts, memory, security, and research-first development for Claude Code, Codex, Opencode, Cursor and beyond.
Atomicmemory
Portable semantic memory for AI agents: core engine, TypeScript SDK, framework adapters, MCP server, CLI, and host plugins.
Install
npm install @atomicmemory/sdkInstall command taken from the project's README or registry entry. Always review code before running it.
README excerpt
Read full README on GitHub →AtomicMemory
Inspectable, correction-aware memory for agents and AI applications.
AtomicMemory gives agents durable context across sessions without coupling your application to one model, framework, or deployment. Start with managed Hosted Cloud, run the open-source Core locally, or integrate through the TypeScript SDK and MCP server.
Documentation · Hosted Cloud · Open-source quickstart · Why inspectable memory matters
Headline Results
AtomicMemory v66 is leading performance/cost on BEAM-100K, BEAM-1M, and LoCoMo10 under matched methodology against published competitors. On BEAM-10M it matches the strongest published Mem0-new result while leaving Hindsight-scale temporal retrieval as the known open frontier.
| Benchmark | AtomicMemory v66 | Position | Cost/Q | Sample |
|---|---|---|---|---|
| BEAM-100K lenient | 0.7375 | Parity with Hindsight at 0.75 | $1.26 | n=80 |
| BEAM-1M lenient | 0.6625 | Leading Performance/Cost; +0.022 vs Mem0 paper | $0.083 | n=80 |
| BEAM-10M lenient | 0.4875 | Parity with Mem0-new at 0.486 | $0.081 | n=80 |
| LoCoMo10 GPT-4o-mini binary | 0.8396 | Leading Performance/Cost; +0.171 vs Mem0 paper | $0.066 | n=1540 |
These results put AtomicMemory at or near the published ceiling in each reported category while preserving the lower-cost operating profile that matters for real applications. Reproducibility artifacts and harness details will be published with the benchmark materials.
Quickstart
Install the am CLI and initialize Hosted Cloud in one guided command:
curl --proto '=https' --tlsv1.2 -fsSL https://get.atomicstrata.ai/install.sh | sh -s -- --init
Plain am init defaults to managed Cloud, selects your project, and saves its
project-bound credential. Managed server keys use a per-installation name such
as am-cli-a1b2c3d4e5f6, so another machine's credential is not rotated. No
Docker or OpenAI key is required.
Non-interactive Cloud automation uses am init --yes --project <cloud-id>;
Local automation must opt in with am init --local --yes.
The installer runs initialization through the verified binary directly. If it updates PATH, open a new terminal before the next commands or use the shell-specific activation command it prints.
Store a preference and retrieve it:
am memory ingest "I prefer aisle seats when flying."
am memory search "seat preference"
Connect the active profile to an agent host when you are ready:
am integrate --yes --host cursor # or claude-code / codex / opencode
am integrate writes the host's user-level MCP configuration. It does not
install a marketplace plugin.
Choose your path
| Path | Best for | Start here |
|---|---|---|
| Hosted Cloud | Managed memory with the fastest setup | Guided installer above, or am init |
| Connected Local | Running open-source Core on your machine | am init --local |
| TypeScript SDK | Server-side application integration against Core | npm install @atomicmemory/sdk |
Agent integration through MCP works with either an active Cloud or Local profile. See the documentation for framework and host-specific guides.
Why AtomicMemory
Similar plugs
Reverse engineer anything with agents, from app behavior down to native binaries.
Fair-code workflow automation platform with native AI capabilities. Combine visual building with custom code, self-host or cloud, 400+ integrations.
Build Agentic workflows, RAG pipelines, with rich AI model and tool support on one collaborative workspace. Deploy on cloud, VPC, or self-hosted, so teams move from prototype…
User-friendly AI Interface (Supports Ollama, OpenAI API, ...)