Security audit
Security checks for vulnerabilities and agentic risk
This skill is a disclosed local self-critique workflow, with the main caution being to cap or disable the debate loop when you are done.
Before installing, confirm the target directory, review or translate the generated Turkish prompt files if needed, and keep max_iterations set to a reasonable number. Only add the heartbeat snippet if you actually want recurring agent self-critique, and set active to false when finished.
The script's echoed setup messages and embedded markdown instructions are written in Turkish throughout, which imposes a specific language on users. Under the policy, language constraints should either be optional for the user or clearly justified as region-specific; neither is present here.
The skill explicitly instructs the agent to alternate personas and continue critique loops 'until you stop them,' but it does not warn users that this can lead to effectively unbounded iterative execution if no stop condition is enforced. In an agent environment, this can waste tokens, time, and compute, and may cause the agent to get stuck in self-perpetuating debate behavior instead of completing the primary task.
