The Agent Skills Directory

[COMMAND_EXECUTION]: The skill references the execution of local Node.js scripts to facilitate state management and content validation.\n
Evidence: node scripts/run-state-check.js sales in SKILL.md\n
Evidence: node skills/write/scripts/lint.js analyze draft.md in SKILL.md\n- [INDIRECT_PROMPT_INJECTION]: The skill is designed to process external, untrusted data (call transcripts), which creates a potential surface for indirect prompt injection.\n
Ingestion points: The instructions suggest using 'Evidence' from real calls pasted by the user as input for the drafting process (SKILL.md).\n
Boundary markers: There are no explicit delimiters or warnings provided to the agent to ignore embedded instructions within the pasted evidence.\n
Capability inventory: The skill references local script execution via Node.js as the primary capability.\n
Sanitization: No sanitization or validation steps are described for the processed transcripts.

sales