The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill provides instructions to install related dependencies using npx skills add from the vendor's repository (PrunaAI/pruna-skills). These are recognized as legitimate resources from the skill's author.
[COMMAND_EXECUTION]: The documentation includes curl command examples for interacting with the api.pruna.ai endpoint. These commands facilitate file uploads and image prediction requests.
[INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied prompts and image references which are interpolated into API calls. There is a potential attack surface for indirect prompt injection common to tools processing external data.
Ingestion points: User-provided prompts and image URLs in SKILL.md (via HTTP POST body).
Boundary markers: Absent; the skill relies on "Prompt craft" instructions rather than technical delimiters.
Capability inventory: Network operations via curl to api.pruna.ai.
Sanitization: Absent; the skill passes user input directly to the API request payload.
[DATA_EXPOSURE]: The skill uses the environment variable ${PRUNA_API_KEY} for API authentication in code examples, following safe practice for secret management by avoiding hardcoded credentials.

p-image-edit