The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted input including video URLs and text prompts, which creates a surface for indirect prompt injection attacks.
Ingestion points: video_url and user instructions processed at runtime in SKILL.md.
Boundary markers: Absent. The skill does not instruct the agent to use delimiters or ignore instructions that might be contained within the remote data.
Capability inventory: Includes file system write access (~/.pixazo/api-key) and network communication via curl, Python requests, and Node.js fetch to gateway.pixazo.ai.
Sanitization: The instructions do not specify any validation or sanitization for external content before it is processed or passed to API endpoints.
[COMMAND_EXECUTION]: The skill provides instructions for the agent to execute shell commands for credential management and API interaction.
Evidence: Includes shell commands for modifying file permissions (chmod 600) on sensitive files and performing network requests using curl.
[CREDENTIALS_UNSAFE]: The skill handles a Pixazo API key by reading it from environment variables, files, or user input.
Evidence: The skill instructs the agent to read and store the PIXAZO_API_KEY in ~/.pixazo/api-key. While this follows standard secret management practices for CLI tools (including restricting permissions with chmod 600), the agent is instructed to request and persist these credentials locally.

video-trim-api