kami
95a96d87a5
The deployment moved from orchestra.service to Docker Compose, but both CLAUDE.md and AUDIT.md still described the systemd unit. A stopped unit plus a stale /usr/local/bin/orchestra reads exactly like a dead service, so document where the live deployment actually is and that deploying now means rebuilding the compose images. Also record what a live check found: the running image predates the B18 auth commit, all six herdrs are unreachable (the sole remaining blocker on live proof for B13-B17), GET /v1/tasks has no authz gate but sits behind a ufw-restricted bind, and ntfy still 403s. Sharpen the herdr-logging note: successes are never logged, so a herdr with no log line is up, not down, and the logs can't confirm one coming back. Probe the ports directly. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01535A3Y8RtkAi8wYuWhtkEd