AI agents call command_completions to retrieve information from Ragmacs without modifying anything — typically the context-gathering step in research, monitoring, and reporting workflows, before the agent takes action elsewhere.
This is a straightforward introspection tool that lists available commands without modifying any state or executing actions. Similar to function_completions and function_documentation on the same server, it retrieves information for discovery purposes. The blast radius of misuse is minimal: an LLM could learn what commands exist, but cannot invoke them or cause harm via this tool alone.
From the tool's definition Tool returns completions matching a prefix—a pure query operation with no side effects. The description indicates it 'Return[s]' matching commands, which is retrieval of metadata about available interactive commands in Emacs.
Attacks that exploit this kind of access
Return all interactive commands matching a prefix. It is categorised as a Read tool in the Ragmacs MCP Server, which means it retrieves data without modifying state.
Register the Ragmacs MCP server in PolicyLayer and add a rule for command_completions: 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 Ragmacs. Nothing to install.
command_completions is a Read tool with low risk. Read-only tools are generally safe to allow by default.
Yes. Add a rate_limit block to the command_completions 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 command_completions. 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.
command_completions is provided by the Ragmacs MCP server (landermkerbey/ragmacs-mcp). PolicyLayer sits as a proxy in front of this server to enforce policies before tool calls reach the server.
Every MCP server has a record like this.
Type a name, get the same breakdown: verified identity, auth posture, risk grade, capabilities, recommended policy.
Teams ship this data inside their own products. See what a licence covers →