The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes performance data from external sources, creating a potential attack surface for indirect instructions. \n- Ingestion points: The skill accepts '性能数据' (performance data) as input in SKILL.md, which may contain untrusted data from web pages or analysis tools. \n- Boundary markers: There are no explicit delimiters or instructions to the agent to disregard instructions embedded within the ingested data. \n- Capability inventory: The skill instructions direct the agent to run the Lighthouse analysis tool and perform file-system modifications such as image compression and code restructuring (SKILL.md). \n- Sanitization: No data validation, filtering, or sanitization processes are defined for the input data.

gate-perf