The Agent Skills Directory

[SAFE]: The skill is designed for project management and QA validation. It uses standard tools (Read, Glob, Grep, Write) to scan the local project environment for launch readiness indicators.
[DATA_EXPOSURE_AND_EXFILTRATION]: While the skill scans for sensitive patterns such as hardcoded credentials or debug flags, it does so to ensure their removal prior to launch. The skill lacks network access and does not attempt to send any data to external servers.
[COMMAND_EXECUTION]: All operations are restricted to the provided agent tools. The skill does not attempt to execute arbitrary shell commands or external scripts.
[INDIRECT_PROMPT_INJECTION]: The skill reads external project files (such as CLAUDE.md and codebase files). While these represent ingestion points for untrusted data, the skill's behavior is limited to string counting and template generation, which minimizes the risk of the agent following instructions embedded in the analyzed data.

launch-checklist