The Agent Skills Directory

[SAFE]: The 'Empirical Validation' skill provides a framework for the agent to verify its own work using local tools. The instructions prohibit the agent from claiming success without verifiable proof.
[COMMAND_EXECUTION]: The skill utilizes the run_command tool to execute standard development tasks such as npm run build and curl for local API testing. These commands are typical for developer workflows and are used here exclusively for verification purposes.
[DATA_EXPOSURE]: The protocol involves writing evidence to local project files (e.g., .gsd/JOURNAL.md). There are no instructions to read or transmit sensitive user credentials or system configurations.

empirical-validation