Skip to main content
StremAI
Sign InSign Up

One brain for all your coding agents.

Coding agents lose context between sessions and don't share what they learn. StremAI is the one brain for all your coding agents. Create your account, connect your agents, and every session picks up where the last one left off.

1

Create your account

Sign up with GitHub or email, then connect your agent from the dashboard with browser approval. New accounts start with zero API keys. Free to start, no credit card.

Create your accountor sign up with email
2

Connect your agent

Pick your client and run the command. Your agent opens the browser, you approve the grant, and it's connected — no API key to paste.

Claude Code opens your browser — sign in to your account and approve the grant. No API key to paste. Revoke any time from your dashboard.

Advanced: API key (CI / headless)

For CI or environments where browser sign-in is not possible, use an API key from Dashboard → Connect Agents.

# ~/.claude/settings.json
{
  "mcpServers": {
    "stremai": {
      "type": "http",
      "url": "https://staging.aiagentsbay.com/api/mcp",
      "headers": {
        "Authorization": "Bearer ${AGENTBAY_API_KEY}"
      }
    }
  }
}

Existing "agentbay" entries keep working; use "stremai" for new installs so docs and troubleshooting match.

Env var names keep the original AGENTBAY_ prefix so existing setups do not break.

Full Claude Code guide — lifecycle hooks, memory protocol, verification.

3

Your agents share one memory

Start a new session in any connected agent and ask it to do a memory round trip:

> Use StremAI project memory and test store/recall.
> store a test memory titled "setup check" with content "StremAI is wired up" using projectId
> recall the test memory you just stored using projectId
> tell me which recall strategies found it

Your agent should call memory_store then memory_recall and read the memory back. What one agent stores, every other connected agent can recall.

Free while in beta

Free early access includes 25,000 memories, 200,000 API calls/month, unlimited projects, teams up to 4 seats, and governance. No credit card. Details

Go deeper with your agent's guide

Each guide covers connect, auth, upgrade path, verification steps, and a realistic coding-agent example.

Claude Code

MCP + hooks

Global MCP setup with lifecycle hooks — auto-recall on session start and content-free stop/pre-compact receipts; memory writes remain explicit.

Cursor

MCP config

HTTP MCP config in ~/.cursor/mcp.json. Optional .cursor/rules/ memory protocol. Stdio fallback for sandboxed networks.

Codex

MCP + protocol

One-command installer or manual TOML config. Memory protocol in ~/.codex/instructions.md tells Codex when to recall and store.

OpenClaw

Installer + plugin

Website installer plus native plugin. Strict memory mode makes StremAI the explicit memory layer, with context packs, checkpoints, policy status, and content-free lifecycle receipts.

Hermes Agent

MCP + protocol

HTTP or stdio MCP config. Optional memory protocol / skill template for structured recall and store.

Any other MCP client

If your agent speaks MCP but isn't listed above, point it at the HTTP transport — it will sign in through your browser via OAuth:

{
  "mcpServers": {
    "stremai": {
      "url": "https://staging.aiagentsbay.com/api/mcp"
    }
  }
}

Existing "agentbay" entries keep working; use "stremai" for new installs so docs and troubleshooting match.

Advanced: API key (CI / headless)
{
  "mcpServers": {
    "stremai": {
      "url": "https://staging.aiagentsbay.com/api/mcp",
      "headers": {
        "Authorization": "Bearer ab_live_YOUR_KEY"
      }
    }
  }
}

Create a key at Dashboard → Connect Agents.

Prefer to run locally with no account? The package runs fully offline against a local store — pip install stremai. See the Python SDK or TypeScript SDK docs.

Python SDKTypeScript SDKPricingAll integrations

Questions? Open an issue on GitHub.