The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructs the agent to install a third-party package @neta-art/cohub-cli from the NPM registry. This package does not originate from a recognized trusted organization or the skill author's established namespace.
[COMMAND_EXECUTION]: The skill performs numerous shell operations using the cohub CLI. It interpolates user-provided inputs and environment variables like <model>, $COHUB_SPACE_ID, and $REFERENCE_AUDIO_URL directly into shell command blocks. This pattern is vulnerable to command injection if the interpolated values contain shell metacharacters.
[INDIRECT_PROMPT_INJECTION]: The skill exhibits an attack surface where it ingests untrusted data from external sources.
Ingestion points: The skill reads model schemas and multimodal data via cohub models show <model> --json and cohub models ls --json (SKILL.md).
Boundary markers: There are no explicit instructions or delimiters to ignore potential instructions embedded within the JSON outputs of the model catalog or schema.
Capability inventory: The skill has the ability to execute shell commands and write to the local file system using the --output flag (SKILL.md).
Sanitization: There is no evidence of sanitization or validation of the JSON data returned by the CLI tool before it is used to influence further agent actions.

cohub-generate