skills/mmx-job-orchestrator/SKILL.md
--- name: mmx-job-orchestrator description: Supervisor for all async MMX jobs (video, music, batch image). Handles: submit → poll → retry → download → clean. Activates when Arif asks to generate video, music, or batch images. Uses mmx exit codes, stderr for progress, and implements agentic retry semantics — not human retry. metadata: {"openclaw": {"emoji": "🎬"}} --- # MMX Job Orchestrator — Async Supervisor for MMX Async MMX jobs (video, music, batch image) are fire-and-forget for humans. For
npx skillsauth add ariffazil/openclaw-workspace skills/mmx-job-orchestratorInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
Async MMX jobs (video, music, batch image) are fire-and-forget for humans. For an ASI agent, they need a supervisor loop: submit → track → retry on failure → download → confirm → clean up.
| Job | Command | Poll Method | Max Retries |
|---|---|---|---|
| Video | mmx video generate --async --prompt "..." | mmx video task get --task-id <id> | 3 |
| Music | mmx music generate --out song.mp3 (sync but slow) | N/A | 2 |
| Image Batch | mmx image generate --prompt "..." --n 3 | N/A | 2 |
| Vision File | mmx vision describe --file-id <id> | N/A | 2 |
# Video — use async to get task-id immediately
mmx video generate --prompt "..." --async
# Returns: { task-id: "..." } or similar — capture the ID
# Music — run in background with timeout
timeout 120 mmx music generate --prompt "..." --out /tmp/song.mp3
# Poll every 30s for up to 10 minutes
mmx video task get --task-id <task-id>
# Check status: pending / processing / complete / failed
Exit code 0 + status complete → proceed to download
Exit code 0 + status pending/processing → re-poll
Exit code non-zero → retry up to max retries
Exit code failure + no progress after max retries → abort, announce failure
# For video
mmx video download --file-id <file-id> --out <destination>
# For music (if sync)
# Already at --out path
[type] ready: <path>| Exit Code | Meaning | Action |
|---|---|---|
| 0 | Success | Proceed |
| 1 | General error | Retry once, then abort |
| 2 | Invalid arguments | Abort — fix prompt |
| 3 | Auth/token error | Abort — token issue |
| 124 | Timeout (from timeout cmd) | Retry once with same prompt |
| Any other non-zero | Unknown error | Retry, log stderr |
Many MMX async jobs stream progress to stderr, not stdout. Capture stderr for diagnostics:
mmx video generate --prompt "..." --async 2>&1 | tee /tmp/mmx-job.log
For batch image jobs:
# Submit all at once (background)
for i in {1..3}; do
mmx image generate --prompt "image $i" --out /tmp/img_$i.png &
done
wait # wait all complete
# Check all exist
for f in /tmp/img_*.png; do
[ -s "$f" ] || echo "MISSING: $f"
done
MMX JOB COMPLETE:
Type: [video/music/image]
Task-ID: [if async]
Output: [path or URL]
Duration: [time from submit to ready]
Retries: [0 if clean, N if retried]
Status: [SUCCESS / FAILED]
MMX JOB FAILED:
Type: [video/music/image]
Task-ID: [if async]
Error: [stderr or exit code]
Retries attempted: [N]
Action: [aborted / escalated to Arif]
development
Governed intelligence skill for AAA as the abstraction, attestation, and abduction control plane across arifOS, APEX, A-FORGE, GEOX, WEALTH, WELL, and the ariffazil profile repository. Use when the user asks to explain or design AAA, route agentic work, reduce chaos/entropy in an arifOS federation task, create AREP/task declarations, classify risk, plan multi-repo changes, review governance boundaries, or translate human intent into evidence-backed, authority-safe, recursively agentic workflows. Provides deterministic F1-F13 floor checking, bounded abduction, and FederationReceipt composition.
development
Check every skill’s “use when” and “do not use when” clauses for collisions, missing negatives, and vague verbs like “help,” “assist,” or “improve.” Load when linting, reviewing, or validating trigger boundaries.
development
Bootstrap, design, and package new skills. Load when capturing user intent for a new skill or drafting its initial instruction framework.
content-media
Diagnose which federation services are up, down, or drifting. Produce a prioritized remediation plan.