media-proxy

SkillMedia

Shared helper that routes ALL paid media generation (FAL image/video, ElevenLabs music) through the GooseWorks proxies so every call bills the Ads agent — never a provider SDK's default host. Host-swaps the FAL queue URLs, loads the agent token from ~/.gooseworks/credentials.json, and returns the result CDN URL. Every video-ad media capability imports this; templates never call a provider directly.

Available today. Use it from your connected AI after setup.

Connect ahel once, and every AI you use reads what you have installed.

Then ask your AI: use the media-proxy skill

What this skill tells your AI

The instructions your AI receives, as published by gooseworks-ai/goose-skills in skills/ads/capabilities/media-proxy/SKILL.md and read by ahel’s review.

The foundation capability for paid media in the video-ad pipeline. It fixes the auth-path conflict where engine scripts called FAL/ElevenLabs directly (billing the wrong account): all paid calls now go through <api_base>/api/internal/{fal-proxy,elevenlabs-proxy} with ?token=&agent_id=, which bills the Ads agent.

Crash-resume (never lose / double-bill a paid render)

A FAL submit BILLS immediately, but the local backend can blip during a multi-minute render. Two built-in protections (automatic for every capability that imports this):

  • Poll-through-outage_fal_run's poll loop re-attaches to the same status/result URL through connection refused / timeout blips instead of crashing.

  • Persist + resume — each submit's request_id + poll URLs are written to ~/.gooseworks/pending-fal-jobs/. If the poller still dies, re-attach instead of re-firing (re-firing double-bills): resume_fal(request_id) in Python, or the CLI:

    resume.py --list                              # resumable (submitted, unfinished) jobs
    resume.py --request-id <id> --out final.mp4   # poll to completion + download
    

    resume_fal NEVER re-submits, so it can't double-charge.

Use it

from media_proxy import fal_generate, fal_generate_video, eleven_music, download

# image (nano-banana / gpt-image / etc.) — inputs must be PUBLIC urls
img = fal_generate("fal-ai/nano-banana/edit",
                   {"prompt": p, "image_urls": [product_url], "aspect_ratio": "9:16"})
# video i2v (kling / seedance / veo)
vid = fal_generate_video("fal-ai/kling-video/v2.1/standard/image-to-video",
                         {"prompt": p, "image_url": keyframe_url, "duration": "10"})
# music bed
eleven_music(prompt, 10500, "music.mp3", force_instrumental=True)

Contracts (load-bearing)

  • Bills the Ads agent?token=&agent_id= from ~/.gooseworks/credentials.json (the CLI writes it; run gooseworks login if missing).
  • Host-swap the FAL queue URLs — submit returns status_url/response_url on queue.fal.run; the helper rewrites them to the proxy base (keeps the path). Never poll queue.fal.run directly (401 + burns credits).
  • FAL inputs that are local files must be PUBLIC urls. The orchestrator hosts a local image/audio via the MCP get_upload_urlget_download_url presigned URL and passes THAT url in. This module does not do MCP uploads (prefer the presigned url; fal-storage-proxy may 404).
  • Only the final *.fal.media url is a real public URL — everything else is behind the proxy.

Related

  • Used by create-image-fal, create-video-fal, create-music-elevenlabs.
  • The goose-video orchestrator hosts local inputs (MCP upload → presign) before calling these.

Signals

GitHub stars
1k
Forks
206
Last commit
Sep 2026
Advanced
Catalog kind
skill
Gateway key
media-proxy
Source
github.com/gooseworks-ai/goose-skills