The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides detailed instructions and examples for executing the genmedia CLI tool to perform model inference and management tasks.
[EXTERNAL_DOWNLOADS]: The run and status commands include a --download flag that automatically fetches files from remote media URLs to the local filesystem.
[DATA_EXFILTRATION]: The upload command enables sending local files to the fal.ai CDN, which is a standard feature for providing inputs to generative models but represents a data movement capability.
[PROMPT_INJECTION]: The skill demonstrates an attack surface for indirect prompt injection as it processes untrusted data from external model outputs.
Ingestion points: Data enters the agent context through CLI command results (e.g., genmedia run, genmedia status) as documented in SKILL.md.
Boundary markers: No specific delimiters or safety instructions are provided to the agent to distinguish between tool output and potential embedded instructions.
Capability inventory: The skill provides capabilities for local file reads and network uploads (genmedia upload), file writes (--download), and arbitrary model execution (genmedia run).
Sanitization: There is no documented sanitization or validation of the structured JSON data returned by the CLI before it is processed by the agent.

genmedia-ref