The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted external data from Craigslist listings, creating an attack surface for indirect prompt injection.
Ingestion points: The skill fetches listing metadata (titles, descriptions, body fingerprints) from Craigslist via the search, postings, and watch command groups.
Boundary markers: The skill body does not define explicit delimiters or instructions to the agent to disregard natural language commands embedded within listing content.
Capability inventory: The skill relies on a CLI that can perform network requests (via the webhook sink) and write to the local filesystem (via profiles and the file sink).
Sanitization: There is no mention of sanitization or filtering logic applied to listing content before it is returned to the agent context.
[EXTERNAL_DOWNLOADS]: The skill documentation instructs users to install external CLI components from the vendor's repository and package registry.
Evidence: Installation steps involve fetching @mvanhorn/printing-press-library via npx and installing binaries from github.com/mvanhorn/printing-press-library using go install.
[COMMAND_EXECUTION]: The skill provides a highly flexible output delivery mechanism that allows routing data to external endpoints or local files.
Evidence: The --deliver flag supports the webhook:<url> scheme, which performs a POST request to a user-specified URL, and the file:<path> scheme for atomic local file writes.

pp-craigslist