orchestrator-v3
PassAudited by ClawScan on May 4, 2026.
Overview
This instruction-only orchestration skill is coherent and has no code or credential use, but users should review its multi-agent plans before allowing impactful work.
This appears safe to install as an instruction-only planning/orchestration skill. Use it for structured project coordination, but approve plans phase by phase, limit sensitive context, and do not allow bulk code changes or deployment work without explicit review.
Findings (2)
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
The agent may propose or coordinate multiple specialist actions, potentially including code changes or deployment steps through other agents.
The skill is designed to route work to specialist agents and coordinate their execution. This is its stated purpose, but it can lead to broad delegated actions if the user approves an overbroad plan.
Automatically route requests to the best specialist based on task analysis.
Review each execution plan and checkpoint carefully, especially before approving large edits, breaking changes, or deployment-related tasks.
Project context, requirements, or other information you provide may be reused across specialist-agent steps or parallel sessions.
The workflow intentionally passes outputs and context between specialist agents. This is normal for an orchestrator, but the artifacts do not define detailed data-boundary rules for what context should or should not be shared.
Step 1: /product-strategist → Define requirements ↓ (output becomes input) Step 2: /backend-architect → Design based on requirements
Avoid including secrets or sensitive data in orchestration context unless necessary, and clarify which files, systems, or information each specialist should access.