The Agent Skills Directory

[SAFE]: The skill implements a robust orchestration methodology with a focus on accountability and verification. It uses local Node.js scripts to manage a machine-readable 'ledger' of tasks and evidence, which is a defensive design against state drift and unverified task completion.
[COMMAND_EXECUTION]: The skill utilizes local Node.js scripts (scripts/ledger-state.mjs, scripts/validate-ledger.mjs) to manage its internal state. These scripts perform standard file I/O operations (reading and writing JSON files) using built-in Node.js modules and do not execute arbitrary shell commands or access sensitive system paths.
[DATA_EXFILTRATION]: No network operations or data exfiltration patterns were detected. The scripts operate exclusively on local files related to task orchestration and do not call external APIs or services.
[PROMPT_INJECTION]: The instructions establish a hierarchy ('root' vs 'worker') and enforce strict rules for evidence submission and validation. This is a security-positive pattern that helps mitigate risks associated with untrusted or hallucinated subagent responses. No malicious bypass or override patterns were found.

orchestrator