run_command
Execute a shell command in the workspace directory. IMPORTANT: This tool requires user approval before execution. The user will see the command and can approve or deny it. Use this for installing packages (npm, pip, brew), running build commands, git operations, or terminal commands. Do not use s...
This record as markdown: /tools/cowork-os/run-command.md
What run_command does on CoWork OS
AI agents invoke run_command to trigger actions in CoWork OS. What it does depends on the arguments the agent supplies, and its effects often reach beyond the immediate call: builds kicked off, notifications sent, workflows started.
Why run_command is rated High
This tool allows arbitrary code execution via shell commands. While it includes a user approval requirement (which provides some protection), the underlying capability is to execute potentially dangerous operations. The severity is high rather than critical because: (1) user approval is required before execution, providing a safety gate; (2) the scope is limited to the workspace directory.
From the tool's definition The tool explicitly executes shell commands via 'Execute a shell command in the workspace directory.' It can run arbitrary commands including package installation, builds, git operations, and terminal commands, which gives it the ability to trigger external…
Attacks that exploit this kind of access
The rule that runs run_command safely
PolicyLayer is an MCP gateway: it sits between your AI agents and CoWork OS, and checks every tool call against a rule you set before the call runs. Nothing changes on the server itself. For run_command, this is the rule to start with:
run_command stays usable, but rate-capped: a runaway agent can't fire it dozens of times a minute. Everything else on the server is denied unless you say otherwise.
The button opens the PolicyLayer dashboard: create your workspace, connect CoWork OS, apply this rule, and every run_command call is checked against it from then on.
Questions about run_command
Execute a shell command in the workspace directory. IMPORTANT: This tool requires user approval before execution. The user will see the command and can approve or deny it. Use this for installing packages (npm, pip, brew), running build commands, git operations, or terminal commands. Do not use shell heredocs or echo/printf redirection to create artifact files when write_file or edit_file is available; use file tools for file creation and editing. It is categorised as a Execute tool in the CoWork OS MCP Server, which means it can trigger actions or run processes. Use rate limits and argument validation.
Register the CoWork OS MCP server in PolicyLayer and add a rule for run_command: allow, deny, rate-limit, or require approval. Point your MCP client at the PolicyLayer proxy URL and the rule is enforced on every call, before it reaches CoWork OS. Nothing to install.
run_command is a Execute tool with high risk. Execute tools should be rate-limited and have argument validation enabled.
Yes. Add a rate_limit block to the run_command rule in your PolicyLayer policy. For example, setting max: 10 and window: 60 limits the tool to 10 calls per minute. Rate limits are tracked per agent session and reset automatically.
Set action: deny in the PolicyLayer policy for run_command. The AI agent will receive a policy violation error and cannot call the tool. You can also include a reason field to explain why the tool is blocked.
run_command is provided by the CoWork OS MCP server (CoWork-OS/CoWork-OS). PolicyLayer sits as a proxy in front of this server to enforce policies before tool calls reach the server.
More on CoWork OS, and thousands of servers like it.
Across the catalogue