The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes the run_shell_command tool and defines an ExecutionProxy Interface (using the rtk prefix) for executing terminal actions such as npm test. These operations are consistent with the framework's stated goal of deterministic execution and test-driven development.
[INDIRECT_PROMPT_INJECTION]: The skill has an attack surface for indirect injection as it ingests data from founder-context.md and mentions handling web content. It possesses potentially dangerous capabilities like run_shell_command and write_file. However, the risk is mitigated by explicit security protocols requiring the redaction of secrets and PII, and instructions to treat all external data as hostile.
[SAFE]: The skill adheres to vendor-specific protocols (Galyarder Labs) and does not contain obfuscated code, unauthorized data exfiltration patterns, or attempts to bypass agent safety guidelines.

pitch-deck