agentleFS
Koushik890Koushik890/higgsfield-api-skill/SKILL.md

higgsfield

Skill1 starsChanged 49 days agoMITView on GitHub

Generate videos and images with the pay-as-you-go Higgsfield API (api.higgsfield.ai) from inside a project - 79 models including Seedance, Kling, Wan, Minimax Hailuo, LTX, Grok Imagine, Soul, Ideogram, Recraft, Qwen Image and Marketing Studio. Price a request, upload reference media, submit, wait, and download to outputs/. Use when the user asks to create or edit a video or image with Higgsfield or any of these models.

  • Reads credentials

10 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use aeon-update, auto-workflow and 10 more.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • aaronjmarsSkill

    aeon-update

    Pull framework updates from the upstream Aeon repo into this instance - 3-way merges canon's new commits into a PR, never clobbering operator config.

    Also in 4 projects that use this

  • aaronjmarsSkill

    auto-workflow

    Two-mode aeon.yml workflow builder - analyze inspects URLs and emits a tiered, signal-verified skill-enablement plan plus an aeon.yml diff; enable flips slugs to enabled:true and opens a PR.

    Also in 4 projects that use this

  • +10 more from aaronjmars

The file

We split this skill into the rules it gives an agent

7 rules. 0 are common, 2 are distinctive, and 5 are very specific to this project.

  • 0

    Common

    Most of the 9 edited versions we read keep these.

  • 2

    Distinctive

    This author's own choices. Worth a look.

  • 5

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    If the user names a model, use it exactly (model id like kling-video-v3.0-pro-text-to-video or its endpoint …

    line 13
  2. Very specific:

    If they don't, suggest one that fits the task (text-to-video, image-to-video, reference-to-video, video-edit, text-to-image, image edit) …

    line 14
  3. Very specific:

    Write JSON to work/higgsfield/<job>.json with only parameters from info.

    line 19
  4. Very specific:

    If it reports no USD price (some Seedance 2.0, Seedance 2.5 reference/edit/extend, Cinema Studio, Genjutsu, Marketing Studio …

    line 27
  5. Distinctive:

    If the estimate itself fails with HTTP 500 (some Kling motion-control and video-edit models), say so.

    line 28
  6. Very specific:

    If run times out: hf.py wait --job <name>, then hf.py download --job <name>.

    line 34
  7. Distinctive:

    Call estimates estimates. The real charge is on the console usage page.

    line 41

“Common” is measured: kept by at least 6 in 10 of the 9 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Higgsfield generation

Helper: python <skill-dir>/scripts/hf.py (Python 3.8+, standard library only). JSON on stdout, progress on stderr. Run it from the user's project folder; jobs are recorded in work/higgsfield/jobs.db, files go to outputs/.

1. Pick the model

  • hf.py models [words] searches all models, e.g. hf.py models kling image. MODELS.md in the skill folder lists them all with example prices.
  • If the user names a model, use it exactly (model id like kling-video-v3.0-pro-text-to-video or its endpoint kling-video/v3.0/pro/text-to-video). Never swap to a different model or variant without asking.
  • If they don't, suggest one that fits the task (text-to-video, image-to-video, reference-to-video, video-edit, text-to-image, image edit) and mention its example price. Default for video: bytedance-seedance-2.5-text-to-video, 5 s, 720p.
  • hf.py info <model> shows every parameter (type, allowed values, default, required), outputs, pricing note, and an example. Read it before writing a request; parameters differ between models, even between variants of one model. For full details open the model's agent_prompt URL from info.

2. Build the request

  • Write JSON to work/higgsfield/<job>.json with only parameters from info. hf.py validate --model M --input F checks it offline.
  • Local images/video/audio: hf.py upload <file> returns a public_url; put that URL in the model's field (image_url, image_urls, video_url, start_image_url, ...). Only upload files the user asked you to use. Supported: jpg, png, webp, gif, wav, mp4.
  • Marketing Studio product shots / ads with a preset: hf.py presets lists preset ids; ask the user to choose; set enhance_prompt: true, preset_id, and the product image in image_urls. Never invent preset ids.
  • hf.py check verifies the key for free. If credentials are missing, tell the user to fill ~/.config/higgsfield/.env (HF_API_KEY_ID, HF_API_KEY_SECRET). Never ask for or print the key, and never put it in commands or project files.

3. Price it

hf.py estimate --model M --input F is free. Tell the user the number and whether it is exact from Higgsfield, computed, or an upper bound.

  • If it reports no USD price (some Seedance 2.0, Seedance 2.5 reference/edit/extend, Cinema Studio, Genjutsu, Marketing Studio flare/sunburst, Minimax H3 reference): show the pricing text and ask whether they accept an unknown price. Only then add --allow-unpriced.
  • If the estimate itself fails with HTTP 500 (some Kling motion-control and video-edit models), say so; ask before submitting without a price.

4. Generate

  • A clear request to generate ("make a 5 s video of ...") authorizes that one generation at the quoted price. Asking for advice, prices, or installing the skill does not.
  • hf.py run --model M --input F --job <name> submits, waits, downloads. Use a short descriptive job name (beach-intro-v1). Add --max-usd N or --budget NAME --budget-usd N only when the user sets a limit.
  • If run times out: hf.py wait --job <name>, then hf.py download --job <name>. Never resubmit to retry.

Protecting the user's money

  • Same --job name is never charged twice. An identical request under a new name is refused unless the user wants another copy (--allow-duplicate).
  • Status unknown after submit means it may have been charged. Do not submit again. Get the request ID from the user (console or Higgsfield support) and run hf.py attach --job <name> --request-id <id> --model <m>.
  • hf.py list shows all jobs with request IDs, status, estimates, files. hf.py cancel --job <name> works only while queued.
  • Call estimates estimates. The real charge is on the console usage page.

5. Check and report

Confirm the files exist and look at them (for video, extract a frame with ffmpeg) before describing the result. Report paths, model, settings, and estimated cost.

Keeping models current

python <skill-dir>/tools/sync_models.py re-downloads every model's agent prompt from Higgsfield and rebuilds models/ and MODELS.md. For one model: python tools/sync_models.py <endpoint>. Project-specific model files in work/higgsfield/models/ override bundled ones.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.