The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests and processes content from an external source (The Movie Database API), which can include attacker-controlled descriptions or metadata.
Ingestion points: Data enters the canister through functions like searchMovies, movieOverview, and nowPlaying in src/backend/main.mo via the tmdb-client package.
Boundary markers: The skill does not define specific prompt delimiters or instructions for the agent to treat the ingested movie data as untrusted content.
Capability inventory: The skill performs outbound HTTPS network requests to TMDb endpoints and writes to the canister's persistent state.
Sanitization: The Motoko implementation uses a toMovie mapping function that enforces a strict schema and provides default values for missing data, though it does not sanitize the underlying text strings for potentially malicious instructions.
[EXTERNAL_DOWNLOADS]: The skill relies on an external library installed through the Motoko package ecosystem.
Dependency: The tmdb-client package (version ~0.1.2).
Installation: The skill instructs the user to execute mops add tmdb-client to retrieve the Motoko bindings from the mops registry.

connector-tmdb