The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted content from the web via search and crawl operations, which is inherent to its primary purpose.
Ingestion points: Data enters the context through anycap search results and Wikipedia markdown retrieved via anycap crawl as defined in references/lookup.md and references/verify.md.
Boundary markers: The skill instructs the agent to request structured JSON and uses sed to strip code fences, providing a level of isolation for data processing.
Capability inventory: The agent uses shell commands (anycap, jq, sed, echo) to process retrieved data and write results to local files.
Sanitization: Relies on structured output prompting to maintain data integrity, though it lacks explicit sanitization for general text retrieved from external sources.
[COMMAND_EXECUTION]: The skill utilizes standard system utilities including jq, sed, and column alongside the anycap CLI. All shell command templates are constructed with proper variable quoting to ensure safe execution during data processing and table generation.
[EXTERNAL_DOWNLOADS]: The skill performs network operations to retrieve player statistics and biographical data from the vendor's own API and authoritative sources like Wikipedia. These operations are transparently documented and consistent with the skill's stated purpose.

anycap-worldcup-predict