READ-ONLY — return the recommended upgrade flow for the running install path. Combines (1) process.argv/process.execPath heuristics that classify the install as one of npm-global / npx / bundled-binary / from-source / unknown with (2) cached state from the once-per-session npm-registry version ch...
AI agents call get_update_command to retrieve information from VaultPilot MCP without modifying anything — typically the context-gathering step in research, monitoring, and reporting workflows, before the agent takes action elsewhere.
Even though get_update_command only reads data, uncontrolled read access leaks sensitive information and racks up API costs — an agent caught in a retry loop can make thousands of calls a minute without anyone noticing.
Attacks that exploit this kind of access
READ-ONLY — return the recommended upgrade flow for the running install path. Combines (1) process.argv/process.execPath heuristics that classify the install as one of npm-global / npx / bundled-binary / from-source / unknown with (2) cached state from the once-per-session npm-registry version check the server already runs lazily on first tool call. Returns: current (server version), latest (most recent npm registry response, or null if the lazy check hasn't resolved yet), updateAvailable (strict-newer comparator), installPath (detected kind), command (the one-liner to run), restartHint (post-upgrade restart note), and an optional note field that flags caveats (unknown install path → defer to INSTALL.md; unresolved version check → can re-run). AGENT BEHAVIOR: call this when the user asks to upgrade, when the VAULTPILOT NOTICE — Update available block appears and the user wants to act on it, or when the user asks 'how do I update vaultpilot-mcp'. Surface command to the user verbatim — do not execute it autonomously. The detection is a heuristic; if installPath is unknown, ask the user which install path they used. Pure local introspection + cache read; no RPC, no fresh network call (the kickoff already did that). Never throws. It is categorised as a Read tool in the VaultPilot MCP MCP Server, which means it retrieves data without modifying state.
Register the VaultPilot MCP server in PolicyLayer and add a rule for get_update_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 VaultPilot MCP. Nothing to install.
get_update_command 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 get_update_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 get_update_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.
get_update_command is provided by the VaultPilot MCP server (vaultpilot-mcp). PolicyLayer sits as a proxy in front of this server to enforce policies before tool calls reach the server.