Generating plain-English description…
Plain-English summary of the active rules — what agents can do, what gets blocked, and what needs a human.
This policy applies only to the `langgraph-prospect-*` agents in the `dev` environment. They may use the tools `read_customer_profile`, `score_icp_fit`, `draft_outreach_email`, and `send_outreach_email`, and they may use the models `gpt-5-mini` and `claude-sonnet-5`; however, `send_outreach_email` requires approval before it can be used. Each request is limited to 3,000 tokens, with warnings sent at 80% usage, and repeated overages are controlled with a hard deny after the 2x overage threshold and a maximum of 3 overage events in a 24-hour UTC window. PII handling is allowed, but prompt injection and data exfiltration attempts should trigger warnings, and any action outside the allowed window should be denied.