USER MANUAL · PRODUCTIVITY

Tools Monitor

Track every tool your agents use, assess risk levels, and catch failures before they cascade.

/tools

Overview#

The Tools Monitor provides a complete inventory of all tools available to your OpenClaw agents. Each tool is cataloged with its risk classification, usage statistics, and failure count — giving you a clear picture of your agent toolchain at a glance.

Beyond inventory, the Tools Monitor enables proactive risk management. Tools are automatically classified by risk level (critical, high, medium, low) based on their capabilities and potential impact. High-risk tools like file system writers or shell executors are flagged so you can review and govern them before they cause problems.

Prerequisites#

Prerequisites

At least one connector synced (see Connectors)
Agents with tools configured in OpenClaw (see Agents)

OpenClaw Configuration#

Tool discovery uses the tools.catalog RPC, which is a per-agent method. ClawButler iterates over each synced agent and calls tools.catalog with the agent's external_id to retrieve the full tool list for that agent.

Tools are defined in the OpenClaw configuration under each agent's tool section. Key points about tool discovery:

-Each agent has its own set of tools — tools.catalog requires an agentId parameter and returns only tools available to that specific agent
-Tools inherit from agents.defaults — if a tool is defined in the defaults section, all agents that don't override it will have access
-Tool metadata includes the tool name, description, and parameter schema — ClawButler uses this to derive risk classification

Expected Results#

After syncing, the Tools Monitor shows:

-Per-agent tool list with tool name, description, and risk classification badge (critical / high / medium / low)
-Usage statistics including total invocations and failure counts per tool
-Risk distribution summary showing how many tools fall into each risk category
-Tool asset ledger with idempotent upsert — re-syncing does not create duplicates

Web Operations#

The Tools Monitor page is accessible from the Productivity section of the sidebar. It shows a unified view of all tools across all agents.

Tool Inventory

Lists all tools grouped by agent. Each tool displays its name, description, and a color-coded risk classification badge. High-risk tools (e.g., file writers, shell executors, API callers with write access) are surfaced prominently. Read-only tools are classified as low risk.

Risk Assessment

Each tool is assigned a risk level based on its capabilities: critical (system-level access, irreversible operations), high (file writes, external API mutations), medium (data reads with broad scope), and low (read-only, scoped queries). Risk factors include parameter types, access scope, and whether the tool can modify external state.

Tool Monitor Summary

Aggregated statistics at the top of the page: total tools discovered, risk distribution breakdown (how many critical / high / medium / low), total failures across all tools, and the number of agents with tools configured.

Refresh

Click the refresh button to re-sync tool catalogs from OpenClaw. This triggers a tools.catalog RPC call for each agent and updates the local tool asset ledger. The refresh is idempotent — existing tools are updated in place, new tools are added, and removed tools are marked as inactive.

Mobile Operations#

The Tools Monitor is available in the More tab of the mobile app. Features include:

-Scrollable tool list with risk level indicators (color-coded badges)
-Tap any tool to view its full details: description, risk classification, usage count, and failure history
-Pull-to-refresh to re-sync tool data from the server

FAQ#

Q: What determines a tool's risk level?

A: Risk level is derived from the tool's capabilities and potential impact. Tools that can write to the file system, execute shell commands, or mutate external APIs are classified as high or critical risk. Tools that only read data or perform scoped queries are classified as medium or low. The classification considers parameter types, access scope, and whether the tool can produce irreversible side effects.

Q: How can I reduce tool risk across my agents?

A: Review tools flagged as critical or high risk in the Tools Monitor. In your OpenClaw configuration, you can restrict tool access per agent — remove tools that an agent doesn't need, limit tool parameters to safe ranges, or replace a broad tool with a more scoped alternative. After reconfiguring, sync the connector to update risk assessments.

Q: A tool shows as 'unknown risk' — what does that mean?

A: Unknown risk means ClawButler could not determine the tool's risk classification from its metadata. This typically happens when the tool lacks a description or parameter schema. Update the tool's definition in your OpenClaw configuration to include proper metadata, then re-sync to get an accurate risk assessment.