The Agent Skills Directory

[SAFE]: The skill implements operational guardrails using platform-native hooks to monitor tool calls. These hooks are designed to prevent context window waste by reminding the agent to delegate diagnostic tasks rather than performing them locally.
[SAFE]: Analysis of the enforcement mechanisms, including the blocking behavior for redundant bash commands, confirms they are strictly internal to the agent's workflow management and do not involve any network calls, credential access, or unauthorized file system operations.

orchestrator-discipline