AI agents invoke auto_recover to trigger actions in Npm Dev. 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.
This tool executes a multi-step automated recovery workflow (health check → recovery → re-validation) which involves starting/stopping/restarting dev server processes. It doesn't simply read state or write data — it actively triggers external process execution. Misuse could restart wrong projects, consume ports, or cause service disruption across multiple projects, warranting high severity.
From the tool's definition 自動復旧を実行(ヘルスチェック→復旧→再検証)— 'auto_recover' triggers automated health-check, recovery, and re-validation sequences; it runs external processes and restarts dev servers as part of the recovery pipeline.
Attacks that exploit this kind of access
MCPサーバーの自動復旧を実行(ヘルスチェック→復旧→再検証). It is categorised as a Execute tool in the Npm Dev MCP Server, which means it can trigger actions or run processes. Use rate limits and argument validation.
Register the Npm Dev MCP server in PolicyLayer and add a rule for auto_recover: 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 Npm Dev. Nothing to install.
auto_recover 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 auto_recover 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 auto_recover. 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.
auto_recover is provided by the Npm Dev MCP server (masamunet/npm-dev-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 →