dreamina-cli-skill
A skill for Dreamina (即梦): Python wrappers around the CLI submit image and video jobs and check results and credits.
- 133 ★ on the repository
- Last commit
Extended access
Why this level
- The official CLI installs through `curl | bash` from an external address.
- Generations spend Dreamina account credits and the login token sits in `~/.dreamina_cli/credential.json`.
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add skills/dreamina-cli-skillDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- The official CLI installs through `curl | bash` from an external address.
- Generations spend Dreamina account credits and the login token sits in `~/.dreamina_cli/credential.json`.
This is third-party code. Review the repository files before installing.
What it does
The skill wraps the Dreamina (Chinese 即梦) command-line client in a thin layer of Python scripts that an agent calls instead of bare `dreamina`. The wrappers validate local paths, normalise parameter names, return stable JSON with `ok`, `command`, `data` or `error` fields and understand `--dry-run`. `scripts/` holds `text2image.py`, `image2image.py`, `image_upscale.py`, `text2video.py`, `image2video.py`, `frames2video.py`, `multiframe2video.py`, `multimodal2video.py`, plus `query_result.py`, `list_task.py`, `user_credit.py` and login and session scripts. Async jobs return a `submit_id` for polling the result. References live in `references/commands.md` and `references/integration.md`.
Who it is for. People who make videos and covers in Dreamina and want an agent or an automated pipeline to submit jobs and collect results.
Good fit when
- An agent generates an image from text at 2k resolution and the parameters are checked first with `--dry-run`.
- A 5-second 720p video is made from a first frame and a prompt.
- The result of an async job is fetched by `submit_id` and downloaded into a folder.
- Credit balance is checked with `user_credit` before a batch of jobs.
Not a fit when
- There is no Dreamina account: sign-in through OAuth Device Flow is required.
- You need a call without Python 3: the wrappers run through `python3`.
- Credits must not be spent without confirmation: costly jobs run with `--dry-run` first.
Example request
Use Dreamina to make a 2k image of a silver ring on a clean background and show the parameters through dry-run first.Limitations
The skill works only with the official `dreamina` installed, which the command `curl -fsSL https://jimeng.jianying.com/cli | bash` provides; read that script before running it. `python3` and an account sign-in are required (`dreamina login`, `--headless` for environments without a browser). State lives in `~/.dreamina_cli/`: `credential.json`, `tasks.db` and logs. Image generation and upscaling need an explicit `resolution_type`, video needs `video_resolution`. On `AigcComplianceConfirmationRequired` first confirm consent in the web interface. The repository states no license.
How to disable. Delete the skill folder from the agent's skill registry. To sign out run `dreamina logout`; local state lives in `~/.dreamina_cli/`.
FAQ
How do I install the skill?
Install the official CLI with the command from the README, then copy or mount the skill directory into the agent's skill registry keeping relative paths. `python3` and `dreamina` are required.
How do I sign in if the agent has no browser?
Run `dreamina login --headless`: it prints `verification_uri`, `user_code` and `device_code`. After web authorisation check the login with `dreamina login checklogin --device_code=<device_code> --poll=30`.
How do I check the environment works?
Run `dreamina user_credit`. If JSON with a balance comes back, sign-in and the local environment are set up.
What do I do with async jobs?
Keep the `submit_id` from the response. Then `query_result.py --submit-id <id>` returns the status and `--download_dir` in the CLI itself downloads files.
Which video models does the README mention?
Seedance 2.0 (including `seedance2.0_vip` and `seedance2.0mini`) and `seedance2.5`. Duration and resolution limits are set for them.
Discussion
No questions or reviews yet.
Related
SkillsA skill that teaches the agent to build narrated videos through narrator-ai-cli: source, template, music, voice, script and composing
A Claude Code skill that turns one sentence into a vertical AI micro-drama, from script and storyboard to video
Skills for an agent that makes video, images, speech and music through the PixVerse CLI using ready pipelines.
A skill for the LibTV platform: from a text brief an agent makes images and video, edits finished clips and downloads the result.
Eleven skills for short dramas: from idea and script to storyboard and ready prompts for the Seedance video model.
Five image-generation skills: Qwen-Image prompts, logos with shared geometry, diary comics, transparent PNGs and process dioramas
README in short
The README explains why a thin wrapper layer exists, how to install Dreamina CLI, sign in and check credits, and lists the directory layout and sample commands for text-to-image, text-to-video, image-to-video and sessions. The conventions section gathers rules: required resolutions, Seedance model limits and handling of the consent error.
README badge
Original SKILL.md
--- name: dreamina-cli description: Use when an agent needs Dreamina（即梦） generation, task querying, account checks, or login/session operations through the packaged Python wrapper scripts around the dreamina CLI. --- # Dreamina CLI Use this skill when you need Dreamina（即梦） image or video generation, task querying, task listing, credit checks, or session management. 即梦 is the Chinese product name of Dreamina. If the user says 即梦, treat it as Dreamina and use this skill. Prefer the packaged Python wrapper scripts in `scripts/` over calling raw `dreamina` directly. The wrappers add path validation, command normalization, structured JSON output, `--dry-run`, and consistent error handling. Treat `dreamina -h` and `dreamina <subcommand> -h` as the final source of truth when a specific flag or supported value is unclear. ## Default workflow 1. If you need capability discovery, run `python3 ./scripts/list_capabilities.py --format json`. 2. Choose the narrowest wrapper script for the user request. 3. For expensive or ambiguous requests, run the wrapper with `--dry-run` first to inspect the exact CLI arguments. 4. Run the real wrapper command. 5. Read the wrapper JSON response instead of relying on shell exit code alone. 6. For async generation, keep the returned `submit_id` and use `query_result.py` if the user wants follow-up polling. ## Script selection - Image generation: - `text2image.py` - `image2image.py` - `image_upscale.py` - Video generation: - `text2video.py` - `image2video.py` - `frames2video.py` - `multiframe2video.py` - `multimodal2video.py` - Task and account operations: - `query_result.py` - `list_task.py` - `user_credit.py` - Session operations: - `login.py` - `login_checklogin.py` - `logout.py` - `relogin.py` - `session.py` - Capability discovery: - `list_capabilities.py` - CLI version check: - `version.py`