skills/b4r7x/nuke-skills/nuke-perf/Gen Agent Trust Hub

nuke-perf

Pass

Audited by Gen Agent Trust Hub on Aug 15, 2026

Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes shell commands (e.g., npm test, pytest, cargo test) to establish performance baselines and validate optimizations. This execution is scoped to user-defined targets and repository manifest scripts, which is consistent with its primary purpose.
  • [DATA_EXPOSURE_AND_EXFILTRATION]: The skill creates and writes to a .nuke/ directory for storing run artifacts such as performance plans, profile data, and measurement logs. This ensures a transparent and reproducible audit trail for all changes.
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests external data from profiler results and command outputs. 1. Ingestion points: Phase 1 measurement and profiling data. 2. Boundary markers: Not specified in instructions. 3. Capability inventory: Shell command execution and file system writes. 4. Sanitization: Quantitative verification against baselines (Phase 4) and mandatory adjudicator role review (Phase 2).
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 15, 2026, 07:40 PM
Security Audit — agent-trust-hub — nuke-perf