The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill documentation specifies the installation of the 'sports-skills' Python package from a public registry as a requirement.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process external Formula 1 data which serves as a potential vector for indirect prompt injection if the source data is untrusted.
Ingestion points: Data enters the agent's context through commands like get_race_results, get_session_data, and get_lap_data within the SKILL.md and references/commands.md files.
Boundary markers: No specific delimiters or safety instructions are provided to the agent to treat the retrieved racing data as untrusted text.
Capability inventory: The skill's capabilities are limited to data retrieval commands; there are no tools for file system modification, network exfiltration, or arbitrary code execution provided.
Sanitization: There is no evidence of sanitization or filtering logic applied to the external data before it is presented to the agent.

fastf1