The Agent Skills Directory

[DATA_EXPOSURE_AND_EXFILTRATION]: The skill performs network operations to an external, non-whitelisted domain to facilitate its core functionality.
Evidence: Network requests are made to https://api.eng0.ai/api/data/images/generate using curl, Python's requests library, and the Node.js fetch API as documented in SKILL.md and README.md.
[COMMAND_EXECUTION]: The documentation provides multiple examples of shell commands to interact with the API.
Evidence: SKILL.md and README.md include several curl command examples for generating images with various parameters.
[INDIRECT_PROMPT_INJECTION]: The skill processes output from an external API that could contain malicious instructions designed to influence the agent.
Ingestion points: The description field returned in the JSON response from api.eng0.ai (documented in SKILL.md).
Boundary markers: Absent. There are no instructions to the agent to treat the API's description field as untrusted data or to ignore potential instructions within it.
Capability inventory: The skill documentation includes Python and Node.js examples that perform file-system writes (Path.write_bytes and fs.writeFileSync) and network requests.
Sanitization: Absent. There is no evidence of filtering, escaping, or validating the description string before it is returned to the agent's context.

nano-banana