Repeated exec approval prompts
User complained about repeated approval popups and inability to disable them as configured.
XMAN does not replace native OpenClaw exec-approval enforcement.
Read the primary sourceApproval fatigue and unchecked automation are opposite failure modes. The goal is a bounded authorization design that makes the right requests visible without treating a prompt as proof.
Historical exec-approval bug reports · Separate XMAN manager policy · Native OpenClaw rules remain in force
User complained about repeated approval popups and inability to disable them as configured.
XMAN does not replace native OpenClaw exec-approval enforcement.
Read the primary sourceMac app user reported prompts for simple commands despite local Always Allow configuration.
Do not imply automatic approval is always safer.
Read the primary sourceReporter described exec prompts recurring despite ask=off, disrupting unattended cron tasks.
Different control surface from XMAN operator approval.
Read the primary sourceOpenClaw exec approval settings remain OpenClaw’s own security boundary. XMAN’s bounded private requester API handles separate manager-governed operations. The routine resource path follows prepare → commit with policy checks and durable receipts; exceptional host scripts require human review through a separate operator channel.
In Balanced mode, routine permitted allocations may proceed within saved constraints. Strict mode deliberately requires review at selected boundaries. The master role does not inherit root or operator authority. XMAN does not promise to remove upstream Mac approval prompts.
Current architecture: trusted Gateway bridge → requester-scoped token → private Unix socket → manager policy/ledger → bounded Podman adapter. A different private operator channel handles reviewed exceptions.
Not shipped: an independent Security Officer daemon that automatically detects and contains every threat.