The Agent Skills Directory

[SAFE]: No security issues were detected. The skill operates entirely through natural language instructions and structured templates.
[COMMAND_EXECUTION]: No shell commands, subprocess invocations, or dynamic context injections (e.g., the !command syntax) were found in the instructions or reference files.
[EXTERNAL_DOWNLOADS]: The skill does not define or request any external packages (Python or Node.js) and does not contain remote script execution patterns like curl|bash.
[DATA_EXFILTRATION]: There are no network operations or calls to external domains. While the skill asks for business metrics, it includes an explicit instruction to avoid requesting sensitive credentials or confidential customer/employee data.
[INDIRECT_PROMPT_INJECTION]: The skill acts as a surface for processing untrusted user data regarding company traction.
Ingestion points: Company snapshot and traction metrics provided during intake (SKILL.md, Step 1).
Boundary markers: The skill relies on natural language structure and instructions to separate assumptions from facts.
Capability inventory: The skill is limited to text generation; it has no file-write, network, or code execution capabilities.
Sanitization: Instructions explicitly mandate using anonymized examples for sensitive data (SKILL.md).

fundraising