// SCAN REPORT

OpenClaw

129 tools. 109 can modify or destroy data without limits.

5 destructive tools with no built-in limits. Policy required.

Last updated: 6 Jun 2026

109 can modify or destroy data

20 read-only

129 tools total

Free to start. No card required.

TOOL MAP

Read (20) Write / Execute (104) Destructive / Financial (5)

WHAT CAN GO WRONG

Destructive tools (action, channels, drop) permanently delete resources. There is no undo. An agent calling these in a retry loop causes irreversible damage.

Write operations (activation, active-memory, agents) modify state. Without rate limits, an agent can make hundreds of changes in seconds — faster than any human can review or revert.

Execute tools (abort, acp, agent) trigger processes with side effects. Builds, notifications, workflows — all fired without throttling.

RECOMMENDED RULES

Deny destructive operations

{
  "action": {
    "deny_if": [
      {
        "conditions": [],
        "on_deny": "Blocked by default. Requires approval."
      }
    ]
  }
}

Destructive tools should never be available to autonomous agents without human approval.

Rate limit write operations

{
  "activation": {
    "limits": [
      {
        "counter": "activation_per_hour",
        "window": "hour",
        "max": 30,
        "scope": "grant"
      }
    ]
  }
}

Prevents bulk unintended modifications from agents caught in loops.

Cap read operations

{
  "ask": {
    "limits": [
      {
        "counter": "ask_per_minute",
        "window": "minute",
        "max": 60,
        "scope": "grant"
      }
    ]
  }
}

Controls API costs and prevents retry loops from exhausting upstream rate limits.

ALL 129 OPENCLAW TOOLS

DESTRUCTIVE 5 tools

Destructive action Destructive channels Destructive drop Destructive reset Destructive uninstall

EXECUTE 17 tools

Execute abort Execute acp Execute agent Execute auth Execute bash Execute capability Execute events_wait Execute gateway Execute infer Execute new Execute node Execute nodes Execute proxy Execute qa Execute restart Execute stop Execute target

WRITE 87 tools

Write activation Write active-memory Write agents Write approvals Write backup Write browser Write canvas Write cap Write card Write channel Write chat Write clawbot Write command Write commitments Write compact Write completion Write config Write configure Write crestodian Write cron Write daemon Write dashboard Write debounce Write debug Write devices Write directory Write dns Write doctor Write dreaming Write echo Write elev Write elevated Write exit Write fast Write gwstatus Write hooks Write host Write id Write input Write instructions Write level Write logs Write mcp Write message Write messages_send Write migrate Write mode Write models Write name Write note Write onboard Write pair Write pairing Write path Write permissions_respond Write phone Write ping Write plugins Write policy Write qr Write queue Write quit Write reasoning Write sandbox Write secrets Write security Write send Write session Write settings Write setup Write skills Write subagents Write system Write tasks Write terminal Write text Write think Write trace Write transcripts Write tui Write update Write usage Write value Write verbose Write webhooks Write wiki Write workboard

FAQ

Can an AI agent delete data through the OpenClaw MCP server? +

Yes. The OpenClaw server exposes 5 destructive tools including action, channels, drop. These permanently remove resources with no undo. PolicyLayer blocks destructive tools by default so they never reach the upstream server.

How do I prevent bulk modifications through OpenClaw? +

The OpenClaw server has 87 write tools including activation, active-memory, agents. Set a rate limit in your policy -- for example, 10 calls per hour prevents an agent from making more than 10 modifications per hour. PolicyLayer enforces this at the gateway, before calls reach OpenClaw.

How many tools does the OpenClaw MCP server expose? +

129 tools across 4 categories: Destructive, Execute, Read, Write. 20 are read-only. 109 can modify, create, or delete data.

How do I enforce a policy on OpenClaw? +

Register the OpenClaw MCP server in PolicyLayer, apply the suggested rules above (adjust the limits to your use case), and point your AI client at the PolicyLayer proxy URL instead of the server directly. Your agents keep the same tools; PolicyLayer evaluates every call against policy before it executes. Nothing to install, live in minutes.

// RELATED SERVERS

Other MCP servers with similar tools.

Starter policies for each. Same risk classification, live on your fleet in minutes.

Enforce policy on every OpenClaw tool call.

Deterministic rules across all 129 OpenClaw tools. Per-identity grants. Full audit log. Live in minutes. Nothing to install.

SECURE OPENCLAW →

Free to start. No card required.

4,600+ MCP servers and 31,000+ tools scanned and risk-classified.

OpenClaw

// TOOL MAP

// WHAT CAN GO WRONG

// RECOMMENDED RULES

// ALL 129 OPENCLAW TOOLS

// FAQ