Runs a focused investigation protocol: formulates 2–4 targeted queries, searches and reads 3–6 credible sources, extracts findings with confidence and source citations, and writes a persistent research document describing actions and recommendations.
Use when you need evidence-backed answers before making decisions — e.g., dependency checks, best-practice discovery, API behavior, or how other projects solved a problem. Not for immediate action or multi-subquestion research (use research-fleet for parallel work).
Suited to agents that can run web searches and web fetch (WebSearch/WebFetch), evaluate source credibility, and write markdown files (Claude, Copilot-like agents, or any agent with web access).
This skill has not been reviewed by our automated audit pipeline yet.
Marshal (Session Commander)
A single-session meta-orchestrator that parses a direction, chains existing skills, and executes multi-step tasks end-to-end without unnecessary user prompts.
Citadel Setup
Interactive first-run wizard that installs hooks, detects project stack, generates harness config, and runs a short live demo to register skills and protect the