Discover success patterns from past task completions. Requires AgentDB for full functionality. Use when running long-horizon goals that should resume automatically across sessions — Claude Code has no native autonomous-loop scheduler. Pair with autopilot_enable + a goal description, then let cron...
AI agents call autopilot_learn to retrieve information from Claude Flow without modifying anything — typically the context-gathering step in research, monitoring, and reporting workflows, before the agent takes action elsewhere.
The tool's primary described action is discovering/reading patterns from historical data, which is a Read operation. However, the broader context (autonomous loop scheduling, cron-based advancement, pairing with autopilot_enable) suggests it may trigger or configure ongoing automated execution. The description is ambiguous about whether it purely reads patterns or also writes/configures autopilot state.
From the tool's definition Discover success patterns from past task completions
Attacks that exploit this kind of access
Discover success patterns from past task completions. Requires AgentDB for full functionality. Use when running long-horizon goals that should resume automatically across sessions — Claude Code has no native autonomous-loop scheduler. Pair with autopilot_enable + a goal description, then let cron fires advance the work. For interactive single-task sessions, native Task is fine. It is categorised as a Read tool in the Claude Flow MCP Server, which means it retrieves data without modifying state.
Register the Claude Flow MCP server in PolicyLayer and add a rule for autopilot_learn: 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 Claude Flow. Nothing to install.
autopilot_learn 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 autopilot_learn 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 autopilot_learn. 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.
autopilot_learn is provided by the Claude Flow MCP server (claude-flow). PolicyLayer sits as a proxy in front of this server to enforce policies before tool calls reach the server.