The Agent Skills Directory

[SAFE]: The skill's primary function is to manage local project files and provide instruction templates for AI agents. All analyzed scripts use standard Node.js built-in modules (node:fs, node:path, node:child_process) to perform their tasks without external dependencies.
[COMMAND_EXECUTION]: The skill includes functionality to execute project-specific verification commands (e.g., npm install, pytest, go test) via a generated init.sh script. This is explicitly documented as the 'Verification' subsystem of the harness and is used to ensure code quality during agent-led development. No unauthorized or hidden command execution patterns were found.
[PROMPT_INJECTION]: While the skill contains extensive instructions for AI agents within its templates (e.g., templates/agents.md), these are functional requirements designed to constrain agent behavior to a specific scope and workflow. They do not attempt to bypass safety filters, extract system prompts, or override global agent behavior maliciously.
[DATA_EXFILTRATION]: No data exfiltration patterns were detected. The skill reads local project files only for the purpose of structural auditing and scoring, and it lacks the network capabilities to transmit data externally.

firstsun-harness