1claw OpenClaw Plugin

OpenClaw gateway plugin for 1claw.

Repository: github.com/1clawAI/1claw-openclaw-plugin
npm: @1claw/openclaw-plugin
Docs: OpenClaw Plugins · 1claw

OpenClaw agents need vault access, transaction signing, and sometimes an inspected LLM path. Running a separate MCP process works, but this plugin registers 27+ native tools inside the gateway itself. Secrets are fetched at runtime. Outbound messages get scanned for leaked values. Optional Shroud routing sends LLM traffic through the TEE proxy when the agent has it enabled.

Install one npm package, set ONECLAW_AGENT_API_KEY, restart the gateway. Your agent gets secrets, signing, automations, memory, runtimes, and directory tools without extra wiring. Toggle features (redaction, injection, Shroud, slash commands) in plugins.entries.1claw.config.features.

---

Features

• Native agent tools — 27+ tools for secrets, vaults, policies, sharing, signing keys, multi-chain transactions, execution intents, automations, memory, runtimes, and discovery (EVM + Bitcoin, Solana, XRP, Cardano, Tron; optional, configurable)
• Automations — Create and manage cron-based scheduled tasks via oneclaw_create_automation, oneclaw_list_automations, oneclaw_delete_automation
• Agent Memory — Store and search persistent vector memory via oneclaw_store_memory, oneclaw_search_memory
• Runtimes — Deploy and manage agent runtime environments via oneclaw_deploy_runtime, oneclaw_list_runtimes
• Discovery — Publish agents to the 1Claw directory via oneclaw_publish_agent, oneclaw_search_directory
• Secret redaction — Scan outbound messages and redact leaked secret values (default on)
• Secret injection — Replace {{1claw:path/to/secret}} placeholders at prompt time (opt-in)
• Shroud routing — Route LLM traffic through Shroud TEE when the agent has shroud_enabled (opt-in)
• Key rotation monitor — Background warnings for secrets expiring within 7 days (opt-in)
• Slash commands — /oneclaw, /oneclaw-list, /oneclaw-rotate, /oneclaw-memory (optional)
• Gateway RPC — 1claw.status for programmatic health/status
• Bundled skill — 1claw skill (skills/1claw/SKILL.md) auto-discovered by OpenClaw

All features are toggled via plugins.entries.1claw.config.features. Auth uses config or env vars.

---

Install

openclaw plugins install @1claw/openclaw-plugin

Or from the repo (e.g. when developing or using as a submodule):

openclaw plugins install -l ./path/to/1claw-openclaw-plugin

---

Config

Minimal config (config file or env):

{
  plugins: {
    entries: {
      "1claw": {
        enabled: true,
        config: {
          apiKey: "ocv_..."
          // agentId, vaultId, baseUrl, shroudUrl optional
          // features: { tools: true, secretRedaction: true, ... }
        }
      }
    }
  }
}

Env fallback: ONECLAW_AGENT_API_KEY, ONECLAW_AGENT_ID, ONECLAW_VAULT_ID, ONECLAW_BASE_URL, ONECLAW_SHROUD_URL.

Restart the OpenClaw Gateway after changing config.

---

Tool names

When enabled, tools are registered with a oneclaw_ prefix (e.g. oneclaw_list_secrets, oneclaw_get_secret). Add them to your agent’s tools.allow (e.g. "1claw" or specific names).

---

Slash commands

Command	Description
/oneclaw	Connection status, vault info, token TTL, features
/oneclaw-list	List secret paths (optional prefix arg)
/oneclaw-rotate	Rotate a secret: /oneclaw-rotate <path> <new-value>

---

Development

npm install
npm run typecheck

• TypeScript only (no build step required for OpenClaw; jiti loads .ts at runtime).
• Optional: npm run build to emit dist/ (not required for openclaw plugins install when using source).

As a submodule in 1claw

From the 1claw repo root:

git submodule add https://github.com/1clawAI/1claw-openclaw-plugin.git packages/openclaw-plugin
git submodule update --init --recursive

Clone 1claw with the submodule:

git clone --recurse-submodules https://github.com/1clawAI/1claw.git

---

License

MIT © 1claw