skills/aelaguiz/arch_skill/arch-step/Gen Agent Trust Hub

arch-step

Pass

Audited by Gen Agent Trust Hub on Jun 16, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [COMMAND_EXECUTION]: The skill utilizes a local Python script (scripts/arch_stage_gate.py) to manage a 'receipt' system within markdown planning documents. This ensures that the agent follows the required planning stages (research, deep-dive, phase-plan, consistency-pass) in the correct order before proceeding to implementation. It also includes several cleanup scripts (scripts/upsert_*.py) intended to remove legacy configuration hooks from the environment.
  • [EXTERNAL_DOWNLOADS]: The external-research command instructs the agent to perform web research for general architectural topics and best practices. The instructions explicitly limit this activity to generalizable topics rather than project-specific trivia to maintain grounding.
  • [DATA_EXPOSURE]: The skill involves reading and writing to repository files, including source code, documentation, and a dedicated 'worklog' file. This is consistent with its stated purpose as an implementation and architectural tool.
  • [SAFE]: The skill implements strict quality controls, such as a 'North Star confirmation' gate, depth-first implementation strategies, and a mandatory implementation audit pass. The code-modification and command-execution behaviors are transparently documented as core components of the workflow.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 16, 2026, 09:15 AM
Security Audit — agent-trust-hub — arch-step