AI agents call estimate_remaining to retrieve information from Headroom without modifying anything — typically the context-gathering step in research, monitoring, and reporting workflows, before the agent takes action elsewhere.
This tool only retrieves and projects current rate-limit and context budget metrics. It reads internal state and performs calculations to return estimates—no data is modified, no commands are executed, and no external systems are affected.
From the tool's definition "Projections from the burn model: % left and minutes to reset per window, projected exhaustion time, context tokens left before compaction."
Attacks that exploit this kind of access
Projections from the burn model: % left and minutes to reset per window, projected exhaustion time, context tokens left before compaction. It is categorised as a Read tool in the Headroom MCP Server, which means it retrieves data without modifying state.
Register the Headroom MCP server in PolicyLayer and add a rule for estimate_remaining: 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 Headroom. Nothing to install.
estimate_remaining 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 estimate_remaining 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 estimate_remaining. 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.
estimate_remaining is provided by the Headroom MCP server (tyejcoleman/headroom). 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 →