The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill implements functions to retrieve content from external network locations.
The directory::skills::download_from_repo function allows pulling skill folders from GitHub repositories.
The directory::skills::download_from_registry function fetches worker bundles from the vendor's registry at api.workers.iii.dev.
[COMMAND_EXECUTION]: The skill performs file system modifications during its installation phase.
The download functions write external worker bundles to the local disk, making them available for the engine to load.
[INDIRECT_PROMPT_INJECTION]: The skill ingests and serves external documentation and metadata, which creates an attack surface for instructions embedded in that data.
Ingestion points: Untrusted data enters the agent context via directory::skills::get (local markdown), directory::registry (API responses), and directory::skills::download_from_repo (GitHub content).
Boundary markers: There are no specified delimiters or instructions for the agent to ignore potentially malicious content within the documentation it reads.
Capability inventory: The skill possesses the ability to write to the file system and perform network requests to external domains.
Sanitization: No sanitization, validation, or integrity checking of the downloaded worker bundles or markdown files is described in the skill logic.

iii-directory