The Agent Skills Directory

[SAFE]: The skill is designed to guide the agent in writing a markdown document (a QC plan). It does not contain any executable code, command-line operations, or network requests. The instruction 'You write the plan. You run nothing, and you write no findings' explicitly restricts the agent's actions to text generation within the defined template.
[SAFE]: The frontmatter includes disable-model-invocation: true, which restricts the agent's ability to trigger other tools or models, further limiting the potential attack surface.
[SAFE]: The 'Explore' and 'Interview' phases involve gathering information about commits, migrations, and environment variables. While this constitutes data ingestion, the skill only uses this information to populate a text document and does not attempt to exfiltrate data or access sensitive file contents like private keys or credential stores.

to-qc