fal-prompting — Claude Code skill: prompt craft for fal.ai models
Holds per-family prompting rules for fal.ai endpoints. Useful when a prompt to a chosen model keeps coming back generic.
npx skills add fal-ai-community/skills --skill fal-prompting -a claude-codeWhat fal-prompting does
The skill tells you how to phrase requests to different model families on fal.ai. Each family has its own habits, and a prompt that lands on one model falls flat on another. Separate references cover Kling, GPT Image 2 and Happy Horse, loaded when the situation calls for them.
Universal principles are in there too. Instead of prestige adjectives, describe what is visible: light, material, lens. Style tags get expanded into concrete visual targets. Change one variable per iteration, or the comparison tells you nothing.
Endpoint calls run through the genmedia CLI, so you need genmedia installed and one genmedia init run. Before relying on a negative prompt, seed or multi-prompt field, check the schema with genmedia schema <endpoint_id> --json — families differ in what they accept. Picking the model itself is a different job, that is fal-models-catalog.
What it helps with
- figure out how to prompt a specific fal.ai model family
- rewrite a generic prompt into visible visual facts
- expand a style tag into concrete visual targets
- compare prompt variants while changing one axis at a time
- check which control fields an endpoint actually supports
Example prompts for the agent
- How do I prompt Kling for this shot?
- My Claude Image 2 prompts keep coming back generic, what should I change?
- Give me the prompt structure for Happy Horse
How to install fal-prompting
1Skills CLI (Vercel)
npx skills add fal-ai-community/skills --skill fal-prompting -a claude-code2Manually: put the files in place
~/.claude/skills/fal-prompting/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/fal-prompting/
1Skills CLI (Vercel)
npx skills add fal-ai-community/skills --skill fal-prompting -a codex2Manually: put the files in place
.agents/skills/fal-prompting/SKILL.mdFor all projects use ~/.agents/skills/
1Skills CLI (Vercel)
npx skills add fal-ai-community/skills --skill fal-prompting -a cursor2Manually: put the files in place
.cursor/skills/fal-prompting/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install fal-ai-community/skills fal-prompting2Manually: put the files in place
.github/skills/fal-prompting/SKILL.md1Skills CLI (Vercel)
npx skills add fal-ai-community/skills --skill fal-prompting -a gemini-cli2Platform CLI
gemini skills install https://github.com/fal-ai-community/skills --path skills/fal-promptingManually: put the files in place
.devin/skills/fal-prompting/SKILL.mdThe legacy .windsurf/skills/ path still works
Skills CLI (Vercel)
npx skills add fal-ai-community/skills --skill fal-prompting -a opencodeClaude.ai
Zip the folder and upload it in Settings → Capabilities → Skills
Step by step, with commands for each platform: How to install skills in Claude Code: five ways
Questions about fal-prompting
How do I install fal-prompting in Claude?
Run npx skills add fal-ai-community/skills --skill fal-prompting -a claude-code. The skill lands at ~/.claude/skills/fal-prompting/SKILL.md.
What does fal-prompting do?
It gives per-family prompting rules for fal.ai models plus universal principles: describe visible facts instead of adjectives, change one variable per iteration.
How is fal-prompting different from fal-models-catalog?
This one covers how to talk to a model you already picked. fal-models-catalog handles endpoint selection.
