Multi-Agent Collaboration System Skills

• Risk level: Low
• Analyzed: Feb 14, 2026, 3:51 PM
• Findings: 0
• Version: 2.0.1

Overview

The skill's files, templates, and runtime instructions match its stated purpose (a multi-agent, document-driven collaboration toolkit); there are no unexplained credential requests, unsafe downloads, or code that contradicts the description.

This package appears to do exactly what it says: provide templates, docs, and small init scripts for a document-driven multi-agent workflow. Before installing or running scripts: 1) Inspect scripts (init.sh and setup.sh) — note setup.sh copies templates into the parent directory (run from the intended location or adjust paths). 2) Do not store secrets in project files (the templates explicitly say 'No secrets'). 3) Be aware the guide encourages spawning sub-agents and cron-style automation — only enable scheduled/autonomous tasks if you trust the agent environment and understand the platform permissions and costs. 4) qmd is recommended for indexing but is not installed by the skill; install/configure it separately if you want on-demand indexing. 5) If you use this in a repo, run it in a sandbox or branch first to confirm file placement and behavior.