herdr-orchestrator

SUSPICIOUS. The skill's core purpose—local multi-agent orchestration in herdr—is coherent and uses official same-org herdr automation patterns plus a documented Claude CLI flag. However, its actual footprint is high risk because it spawns autonomous worker agents with permission checks disabled, can auto-approve prompts, runs verification commands, and can close panes/workspaces. There is no clear credential exfiltration or third-party credential forwarding, so this is not confirmed malware, but the approval-bypass behavior is disproportionate and materially weakens safety controls.