Generate AI images, videos, and audio via the Picsart gen-ai CLI.
npx skills add https://github.com/PicsArt/gen-ai-skills --skill gen-ai-use-picsart
Please help me install this Agent Skill. Skill: gen-ai-use Source: https://github.com/PicsArt/gen-ai-skills/tree/main/skills/gen-ai-use Command: npx skills add https://github.com/PicsArt/gen-ai-skills --skill gen-ai-use-picsart
gen-ai CLI correctly — from authentication and single generations to batch runs, Drive uploads, and troubleshooting — without guessing at flags or stale model names. ## Core Features & Use Cases - Generation workflows: Run text-to-image, image-to-image, text-to-video, and image-to-video generations with correct model IDs, aspect ratios, and input mappings (e.g. -i auto-mapped to startFrame for i2v models). - Batch and Drive operations: Execute manifest-based batch runs with concurrency control and resume failed jobs, plus upload, download, and list files in Picsart Drive. - Diagnostics and scripting: Use --dry-run, --debug, --script, and --no-input for CI-safe, pipe-friendly automation, with a troubleshooting table for common errors. - Use Case: A user asks to generate 20 product image variants overnight. The agent writes a JSON manifest, validates it with --dry-run, runs gen-ai batch run -c 20, then summarizes results.json and retries failures with batch resume. ## Quick Start Ask the agent to generate an image with the gen-ai CLI, for example: "Use gen-ai to generate a sunset over mountains with the flux-2-pro model and save it to my Drive."High-intent search queries and answers about installing and using this skill.
Run `gen-ai generate -m <model-id> -p "your prompt" -s` for a fully specified non-interactive generation. Check available model IDs first with `gen-ai models`, since IDs change as new versions ship.
Write a JSON manifest with a `jobs` array where each job has a unique `id`, `model`, and `prompt`, then run `gen-ai batch run manifest.json -c <concurrency> -o <dir>`. Validate first with `--dry-run` and retry failures using `gen-ai batch resume <dir>/results.json`.
Pass `--script` (shorthand for `--silent --quiet --json`) for clean pipe output, and add `--no-input` so the CLI fails fast instead of hanging on prompts. For headless auth, set the PICSART_ACCESS_TOKEN and PICSART_USER_ID environment variables.
Yes. Pass an input image with `-i` to an i2v model such as VEO, Kling, or Luma, and the CLI auto-maps it to the model's startFrame parameter. Confirm a model's accepted inputs with `gen-ai models info <id>` before running.
`gen-ai whoami` only reads the cached credentials file and does not verify token expiry or make a network call. If a command fails with an AuthError, run `gen-ai login` again; if it fails with a NetworkError, the problem is connectivity, not credentials.
Avoid it for SDK-level integrations, where the `@picsart/ai-sdk` reference is the right source, and for Picsart web or mobile app flows. Models not yet shipped in the SDK catalog are also unavailable to the CLI.
genmedia-labsRuns RunComfy image, video, and audio models from the command line via a single CLI.
SMOrchestra-aiGenerate multi-channel campaign assets from positioning documents and wedges.
Merit-SystemsUploads local files to StableUpload and retrieves public URLs for sharing via agentcash MCP.
dawsonblockManage MCP servers and tools via the mcporter CLI.