agentleFS
fal-ai-communityfal-ai-community/genmedia-cli/.agents/skills/genmedia/SKILL.md

genmedia

Skill82 starsChanged 6 months agoMITView on GitHub

Run a fal.ai model end-to-end with the genmedia CLI. Use this when the user asks to generate an image, video, or audio; convert media; upscale or restyle; run any fal.ai model; or "use genmedia" for a task. Guides discovery, schema inspection, input preparation, execution, and result handling.

4 other projects use a version of this.

Adopted by

Most recently taken up by

  • Other versions4 projects use a version of this.

    Word for word · 0

    None found.

    Edited · 0

    None found.

    Wrote their own version · 4

    1. fal-ai-community/skills★ 265same name
    2. rbrown101010/codex-marketing-skills★ 95same name
    3. rbrown101010/Rileys-Top-Skills★ 9same name
    4. Rajveerx11/skills★ 1same name

    How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.

  • In this repository4 other files it gives its agents.

The file

We split this skill into the rules it gives an agent

11 rules. 0 are common, 2 are distinctive, and 9 are very specific to this project.

  • 0

    Common

    Most of the 4 edited versions we read keep these.

  • 2

    Distinctive

    This author's own choices. Worth a look.

  • 9

    Very specific

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

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

    Use this skill when the user wants to execute a fal.ai model — either by task description (e.g.

    line 13
  2. Distinctive:

    Read all required fields before running.

    line 29
  3. Very specific:

    Use the returned url as the parameter value.

    line 35
  4. Very specific:

    Use {index}, {name}, {ext}, {request_id} placeholders in the template when the model returns multiple files (images[], image_urls[], etc.) …

    line 57
  5. Very specific:

    Which to pass: Use --vector_id if you have it (anything in the library or a search result has one).

    line 81
  6. Very specific:

    Use these flags on every command that takes an asset target: assets favorite, assets unfavorite, assets tags assign / unassign / set, …

    line 85
  7. Very specific:

    Use assets upload only for local files or non-fal URLs — media that isn't already in fal's system.

    line 92
  8. Distinctive:

    When a command exits non-zero, it prints a JSON error object to stderr:

    line 128
  9. Very specific:

    Always use --json so output is machine-readable.

    line 160
  10. Very specific:

    Run genmedia pricing <endpoint_id> --json first if cost is a concern.

    line 161
  11. Very specific:

    If unsure which model to pick, run genmedia docs "<task>" --json for guidance from fal.ai documentation.

    line 162

“Common” is measured: kept by at least 6 in 10 of the 4 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

genmedia workflow

Use this skill when the user wants to execute a fal.ai model — either by task description (e.g. "generate a video of a dog running") or by a specific endpoint_id (e.g. fal-ai/flux/dev). Load genmedia-ref alongside this skill for the full command reference.

Steps

  1. Discover — If no endpoint_id is given, search for a suitable model:
    genmedia models "<task>" --json
  2. Inspect — Get the model's input parameters:
    genmedia schema <endpoint_id> --json
    Read all required fields before running.
  3. Upload files (if inputs include images/video/audio):
    genmedia upload <local_file_or_url> --json
    Use the returned url as the parameter value.
  4. Run the model:
    • Fast model (completes in seconds):
      genmedia run <endpoint_id> --<param> <value> ... --json
    • Slow model (video generation, large jobs):
      genmedia run <endpoint_id> --<param> <value> ... --async --json
      genmedia status <endpoint_id> <request_id> --result --json
  5. Save outputs — when the user expects files on disk, add --download to run or status. The CLI writes every media URL from the result to the filesystem and returns the local paths in downloaded_files[]. Do not curl the URLs yourself; use the flag.
    genmedia run fal-ai/flux/dev --prompt "a cat" --download --json                            # cwd, source file names
    genmedia run fal-ai/flux/dev --prompt "a cat" --num_images 3 --download "./out/{index}.{ext}" --json
    genmedia status <endpoint_id> <request_id> --download ./out/ --json                        # implies --result
    Use {index}, {name}, {ext}, {request_id} placeholders in the template when the model returns multiple files (images[], image_urls[], etc.) to avoid filename collisions. A trailing / or an existing directory path saves files under that directory using their source names.
  6. Return the result to the user. If --download was used, reference the paths from downloaded_files[]; otherwise present the URLs from result clearly.

Assets library (genmedia assets ...)

The fal Assets API lets you persist generated media, organize it into collections, and define reusable characters.

Two ways to identify an asset

  • vector_id — present for any media in the user's library or in a semantic-search result. Returned by assets browse / assets get / assets upload. Used both as a flag on mutations and as the positional argument for assets get and assets tags for-asset.
  • request_id — the generation the user just ran. Returned by genmedia run and genmedia status.

Which to pass. Use --vector_id if you have it (anything in the library or a search result has one); otherwise use --request_id for a fresh generation.

Use these flags on every command that takes an asset target: assets favorite, assets unfavorite, assets tags assign / unassign / set, assets collections add / remove, assets characters create --reference_image <id>.

assets upload is for external media only

Use assets upload only for local files or non-fal URLs — media that isn't already in fal's system. For anything that came out of a genmedia run, use the request_id instead.

Cheatsheet

What you haveUse
vector_id (from assets browse / get / upload)--vector_id <id>
request_id (from genmedia run / status)--request_id <id>
Local file or non-fal URLassets upload <path_or_url>

Example — character from a fresh generation

RUN=$(genmedia run fal-ai/flux/schnell --prompt "an elegant black cat" --json)
REQ=$(echo "$RUN" | jq -r .request_id)

genmedia assets characters create "Whiskers" \
  --description "An elegant black cat sitting on a moss-covered rock" \
  --reference_image "$REQ" \
  --identifier whiskers \
  --json

Same pattern for assets collections add --request_id <id> and assets favorite --request_id <id>. Newly-referenced media — whether by request_id or by a vector_id from a fresh upload — may take a moment to appear in assets browse / semantic search while the embedding finishes.

For the full command tree, run genmedia assets --help or any subcommand with --help.

Handling errors

When a command exits non-zero, it prints a JSON error object to stderr:

{
  "error": "Validation error — num_images: Input should be less than or equal to 4",
  "details": {
    "endpoint_id": "fal-ai/flux/schnell",
    "request_id": "019d...",
    "status": 422,
    "error_type": "ValidationError",
    "validation_errors": [
      { "field": "num_images", "message": "Input should be less than or equal to 4", "type": "less_than_equal", "input": 20 }
    ],
    "body": { "detail": [ ... raw FastAPI payload ... ] }
  }
}
  • status: 422 / error_type: "ValidationError" means the inputs violated the model schema. Read details.validation_errors — each entry names the field, a human message, the validator type, and the input that was rejected. Re-run genmedia schema <endpoint_id> --json if you need the allowed values, fix the offending args, and retry.
  • Other statuses (401 auth, 403 forbidden, 404 endpoint not found, 429 rate limited, 5xx upstream) surface the server message directly in error and do not contain validation_errors. Do not retry 4xx errors blindly — correct the request first.
  • For in-flight model failures, details.logs contains the most recent model-side log lines.

Notes

  • Always use --json so output is machine-readable.
  • Run genmedia pricing <endpoint_id> --json first if cost is a concern.
  • If unsure which model to pick, run genmedia docs "<task>" --json for guidance from fal.ai documentation.

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.