The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructs the agent to install external software packages and binaries from the author's NPM registry and GitHub repository (@mvanhorn/printing-press-library and github.com/mvanhorn/printing-press-library).
[REMOTE_CODE_EXECUTION]: Provides instructions to download and execute code at runtime via go install and npx, targeting remote repositories and package registries.
[CREDENTIALS_UNSAFE]: Accesses and manages sensitive authentication data stored in credentials.toml and cookies files. It also includes functionality to import session data from local browser profiles using the --chrome flag.
[DATA_EXFILTRATION]: The tool supports a --deliver webhook:<url> flag that allows routing command output to arbitrary external URLs. It also contains an optional feedback mechanism capable of sending data to a remote endpoint if VAGARO_FEEDBACK_ENDPOINT is configured.
[PROMPT_INJECTION]:
[METADATA_POISONING]: There is a discrepancy between the metadata author field ('Trevin Chow') and the vendor/owner of the software resources ('mvanhorn').
[INDIRECT_PROMPT_INJECTION]:
Ingestion points: The skill fetches business listings, service menus, and user reviews from the Vagaro marketplace API and website (SKILL.md).
Boundary markers: The instructions do not specify the use of delimiters or 'ignore' warnings for the processing of external marketplace data.
Capability inventory: The CLI possesses capabilities for file system modification (--deliver file), network operations (webhook), and local binary execution.
Sanitization: There is no evidence of filtering or sanitization performed on the untrusted marketplace content before it is processed by the agent.

pp-vagaro