The Agent Skills Directory

[PROMPT_INJECTION]: The instructions focus purely on meeting analysis tasks and do not contain any patterns associated with prompt injection, safety bypasses, or instruction extraction.- [DATA_EXFILTRATION]: No network operations or hardcoded credentials were identified. The use of 'Reference/colleagues.json' is restricted to local name verification as per the skill's purpose.- [COMMAND_EXECUTION]: No shell commands, subprocess calls, or scripts are present in the skill instructions.- [INDIRECT_PROMPT_INJECTION]: The skill contains a surface for indirect prompt injection as it is designed to process external meeting transcripts. Ingestion points: meeting transcripts and notes processed in 'SKILL.md'. Boundary markers: none identified. Capability inventory: none (the skill has no network, file-writing, or command execution permissions). Sanitization: none. The risk is minimal because the skill lacks the capabilities to perform dangerous actions based on injected instructions.

analyzing-meetings