nuke-implement
Pass
Audited by Gen Agent Trust Hub on Jul 27, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [SAFE]: The skill implements an 'asymmetric verification' model where work produced by lower-tier 'worker' agents is independently verified by higher-tier 'session' or 'top' agents in fresh contexts. This architectural design minimizes anchoring bias and ensures that implementation quality is bounded by the stronger validator rather than the weaker generator.
- [COMMAND_EXECUTION]: The skill utilizes standard project-specific tools (e.g., npm test, vitest, pytest, mypy, go test, cargo check) to run 'gates' for testing, type-checking, and linting. These executions are restricted to the commands defined in the repository's own manifests or standard language fallbacks as documented in stack-adapters.md.
- [PROMPT_INJECTION]: The skill includes a dedicated 'security' lens for AI-specific surfaces that explicitly hunts for prompt injection paths, lack of iteration/cost caps, and untrusted content reaching model prompts. This demonstrates a proactive defensive posture against indirect prompt injection and autonomous agent risks.
- [DATA_EXFILTRATION]: No patterns of unauthorized data access or exfiltration were found. The skill's instructions specifically mandate the detection of hardcoded secrets, keys, and tokens as part of its internal auditing process, categorizing them as critical defects.
Audit Metadata