poyo-flux-2-api_skill

Overview

This skill helps prepare, submit, and track PoYo Flux 2 jobs via the https://api.poyo.ai/api/generate/submit endpoint. It guides model selection, builds PoYo-compatible payloads, and describes polling or webhook follow-up to retrieve results. The goal is clear task_id reporting and minimal back-and-forth when generating or editing media with Flux 2 variants.

How this skill works

You pick the correct model id (flux-2-pro, flux-2-pro-edit, flux-2-flex, flux-2-flex-edit), construct a JSON body with a top-level model and an input block, then POST to the submit endpoint with Authorization: Bearer <POYO_API_KEY>. The endpoint returns a task_id; you either poll the unified task status endpoint or wait for a configured callback_url to receive completion notifications. The skill enforces prompt presence, optional reference image handling, and size/resolution choices for intended outputs.

When to use it

• Generate high-quality images or videos where premium fidelity is required (flux-2-pro).
• Edit or modify existing images provided as reference/source (flux-2-pro-edit, flux-2-flex-edit).
• General creative generation with cost/throughput tradeoffs (flux-2-flex).
• When you need a ready-to-submit PoYo payload and clear next steps after submission.
• When you must reliably capture and report the returned task_id for later polling.

Best practices

• Always set top-level model and include input.prompt unless a full payload is supplied.
• Use input.image_urls only for reference/source images; keep prompt focused on desired outcome.
• Specify input.size to match aspect ratio or canvas and input.resolution to balance cost vs quality.
• Store the returned task_id immediately and provide instructions for polling or webhook handling.
• Keep prompts concrete, outcome-focused, and avoid vague adjectives that confuse deterministic rendering.

Example use cases

• Submit a premium image generation job with flux-2-pro and request 2048x1152 resolution for a banner.
• Edit a product photo using flux-2-flex-edit with input.image_urls and a brief edit prompt.
• Prepare a lightweight concept art batch using flux-2-flex to control cost and throughput.
• Create a single-shot retouch job via flux-2-pro-edit and capture the task_id for automated QA polling.