The Agent Skills Directory

[COMMAND_EXECUTION]: The examples/run.sh script contains standard shell commands (curl, jq, sleep) used for testing the API. These operations are limited to the specific service domain (revid.ai) and do not interact with sensitive system files or perform unauthorized actions.
[CREDENTIALS_UNSAFE]: The skill correctly manages authentication by requiring an environment variable REVID_API_KEY. It does not contain hardcoded keys; rather, it provides a safe pattern for users to provide their own credentials via standard configuration or environment variables.
[EXTERNAL_DOWNLOADS]: The skill references and interacts with https://www.revid.ai, which is the official domain for the Revid service as defined by the vendor context. These network operations are necessary for the primary function of the skill (generating and checking video status) and do not involve downloading or executing untrusted code.
[INDIRECT_PROMPT_INJECTION]: The skill takes a user-provided prompt and interpolates it into an API request. While this is an ingestion point for untrusted data, the data is sent to a structured API endpoint specifically designed for video generation, minimizing the risk of the agent itself being influenced by malicious instructions within that text.

revid-prompt-to-video