The Agent Skills Directory

[SAFE]: The skill is designed for integration with the imini Open Platform (imini.ai). All network interactions are restricted to official vendor domains, including openapi.imini.ai for generation tasks and docs.imini.ai for model catalog information.
[SAFE]: Authentication security is prioritized by the skill. Both documentation and scripts (_imini_common.py) enforce the use of environment variables (IMINI_API_KEY) for secret management and explicitly warn against exposing API keys in terminal history or chat transcripts.
[SAFE]: The skill includes bundled scripts (generate_image.py, generate_video.py) to manage task submission and result polling. The practice of reading local user-provided files to encode them as Base64 data URLs is a core functional requirement for providing reference assets to the generation models and is performed transparently.
[INDIRECT_PROMPT_INJECTION]: The skill presents a potential surface for indirect prompt injection as it processes user-provided prompts and asset paths.
Ingestion points: User prompts and model configuration parameters supplied as command-line arguments to the bundled scripts.
Boundary markers: Absent; the skill does not use specific delimiters to isolate user-provided text from internal API logic.
Capability inventory: The skill can perform network requests (urllib.request), read local files (Path.read_bytes), and write results to the filesystem (Path.write_bytes) via the _imini_common.py helper module.
Sanitization: Present; the skill uses argparse for parameter validation and standard json.dumps for API payload serialization, ensuring that untrusted data is not interpreted as executable logic.

imini-generate