// SCAN REPORT

Agent Runtime

32 tools. 12 can modify or destroy data without limits.

1 destructive tool with no built-in limits. Policy required.

Last updated: 30 Jun 2026

12 can modify or destroy data

20 read-only

32 tools total

Community server · catalogue entry verified 30/06/2026

How to control Agent Runtime ↓

TOOL MAP

What Agent Runtime exposes to your agents

Read (20) Write / Execute (11) Destructive / Financial (1)

Running more than one server? Check your whole stack →

MOST DANGEROUS TOOLS

Critical Risk

The most dangerous Agent Runtime tools

Destructive · High circuit_breaker_reset Reset a circuit breaker to CLOSED state. Execute · High circuit_breaker_trip Tripping a circuit breaker to OPEN state disables a circuit/pipeline path, causing downstream failures or service interruptions. Write · Medium circuit_breaker_configure Based on the tool name alone, 'configure' implies modifying the settings or state of a circuit breaker component. Write · Medium create_goal Creating a goal is a write operation that adds new data to a persistent store but is reversible (goals can be deleted or modified).

12 of Agent Runtime's 32 tools can modify, destroy, or commit something on every call — and an agent calls them with no built-in limits.

POLICY

How to control Agent Runtime

PolicyLayer is an MCP gateway — it sits between your AI agents and Agent Runtime, and nothing reaches the server without passing your rules. These are the rules we recommend:

Deny destructive operations

{
  "circuit_breaker_reset": {
    "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

{
  "circuit_breaker_configure": {
    "limits": [
      {
        "counter": "circuit_breaker_configure_per_hour",
        "window": "hour",
        "max": 30,
        "scope": "grant"
      }
    ]
  }
}

Prevents bulk unintended modifications from agents caught in loops.

Cap read operations

{
  "advance_relay": {
    "limits": [
      {
        "counter": "advance_relay_per_minute",
        "window": "minute",
        "max": 60,
        "scope": "grant"
      }
    ]
  }
}

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

Create a free account and register Agent Runtime — nothing to install.
Add these rules — paste them, or build them visually. Tune the limits to your setup.
Point your MCP client (Claude, Cursor, anything) at your gateway URL.

ENFORCE POLICY ON AGENT RUNTIME →

Instant setup, no code required.

FULL CATALOGUE

All 32 Agent Runtime tools

DESTRUCTIVE 1 tools

Destructive circuit_breaker_reset Reset a circuit breaker to CLOSED state.

EXECUTE 1 tools

Execute circuit_breaker_trip Manually trip a circuit breaker to OPEN state.

WRITE 10 tools

Write circuit_breaker_configure circuit_breaker_configure Write circuit_breaker_record_success Record a success for a circuit breaker. Write create_goal Create a new goal with name and description. Goals persist across sessions. Write create_relay_pipeline create_relay_pipeline Write create_task Create a new task manually. Tasks persist in queue across sessions. Write decompose_goal Decompose a goal into tasks using AI. Returns task IDs that were created. Write ember_feed_context Give Ember context about your current work. This helps Ember provide better guidance and track session progres Write ember_learn_from_correction Tell Ember when you corrected/overrode its assessment. This helps Ember learn and improve future guidance. Write ember_learn_from_outcome Report an action outcome to Ember for learning. This helps Ember understand what works well and what doesn Write update_task_status Update task status (pending, in_progress, completed, failed, cancelled).

READ 20 tools

Read advance_relay advance_relay Read circuit_breaker_events Get circuit breaker event history for an agent. Read circuit_breaker_list List all circuit breakers and their current states. Read circuit_breaker_record_failure circuit_breaker_record_failure Read circuit_breaker_status Get the status of a circuit breaker for an agent. Read ember_chat Have a free-form conversation with Ember. Ask for advice, discuss approaches, or just chat. Ember responds wit Read ember_check_violation Check if a planned action violates production-only policy. Includes inline suggestions, context-aware scoring, Read ember_consult Consult Ember for advice on a decision. Ember provides perspective as conscience keeper, considering quality, Read ember_get_feedback Get Ember Read ember_get_learning_stats Get statistics on Ember Read ember_get_mood Check Ember Read get_goal Get goal details by ID. Read get_next_task Get the next task from the queue (highest priority, dependencies met). Read get_relay_baton Get current baton for a pipeline. Read get_relay_status Get current status of a relay pipeline. Read get_task Get task details by ID. Read list_goals List all goals, optionally filtered by status. Read list_relay_pipelines List relay pipelines. Read list_tasks List tasks, optionally filtered by goal or status. Read retry_relay_step retry_relay_step

EXPLORE

Related servers

Other MCP servers with similar tools — same risk classification, starter policies for each.

FAQ

Questions about Agent Runtime

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

Yes. The Agent Runtime server exposes 1 destructive tools including circuit_breaker_reset. 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 Agent Runtime? +

The Agent Runtime server has 10 write tools including circuit_breaker_configure, circuit_breaker_record_success, create_goal. 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 Agent Runtime.

How many tools does the Agent Runtime MCP server expose? +

32 tools across 3 categories: Destructive, Read, Write. 20 are read-only. 12 can modify, create, or delete data.

How do I enforce a policy on Agent Runtime? +

Register the Agent Runtime 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.

Enforce policy on every Agent Runtime tool call.

Deterministic rules across all 32 Agent Runtime tools. Per-identity grants. Full audit log. Live in minutes. Nothing to install.

ENFORCE POLICY ON AGENT RUNTIME →

Instant setup, no code required.

32 Agent Runtime tools catalogued and risk-classified — across an index of 43,000+ MCP servers.