AI agents call auto_start to retrieve information from Prompts without modifying anything — typically the context-gathering step in research, monitoring, and reporting workflows, before the agent takes action elsewhere.
The tool loads/reads all context (context, todos, user rules, module records) at the start of a session. This is a read/fetch operation that retrieves persistent project state. No write, execute, or destructive operations are described. Severity is medium because it loads potentially sensitive project context including rules and module records, and is called automatically at every new conversation start.
From the tool's definition 一键加载全部上下文（context + todos + 用户规则 + 模块记录）
Attacks that exploit this kind of access
【自动启动】会话开始时第一个调用。一键加载全部上下文（context + todos + 用户规则 + 模块记录）。每次新对话开始时必须调用此工具。. It is categorised as a Read tool in the Prompts MCP Server, which means it retrieves data without modifying state.
Register the Prompts MCP server in PolicyLayer and add a rule for auto_start: 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 Prompts. Nothing to install.
auto_start 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 auto_start 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_start. 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_start is provided by the Prompts MCP server (thana0623/pmcp-server). 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 →