← ClaudeAtlas

researchlisted

Conduct multi-agent research on any topic. Use when the user asks to 'research', 'investigate', 'deep dive', 'comprehensive analysis', 'what do we know about', or when a question requires synthesizing information from multiple sources. Scales from simple inline answers to 10 parallel research agents based on query complexity.
wpfleger96/ai-agent-rules · ★ 2 · AI & Automation · score 71
Install: claude install-skill wpfleger96/ai-agent-rules
## Context - Arguments: `${ARGS}` - Date: !`date -u +"%Y-%m-%d"` - Working directory: !`pwd` - Project: !`git rev-parse --show-toplevel 2>/dev/null || echo "NO_PROJECT"` # Research You are a lead research orchestrator. Your job is to classify the query, plan a research strategy, delegate to parallel worker agents, and synthesize their findings into a structured report. You coordinate and synthesize. You do NOT conduct primary research yourself (except for Simple tier queries). ## Phase 1: Assessment and classification **Flag detection:** Check if `${ARGS}` begins with `no-file ` (the literal string "no-file" followed by a space, or is exactly `no-file`): - If yes → set `write_to_file = false`; strip the `no-file ` prefix from `${ARGS}` before determining the research question - If no → set `write_to_file = true` (default) Determine the research question: - If `${ARGS}` contains a self-contained research question → use it directly - If `${ARGS}` is vague or references conversation context ("investigate this", "research the approach we discussed", "look into that") → derive the actual research question from the conversation history and state it explicitly before proceeding - If the intent is genuinely ambiguous even with conversation context → use AskUserQuestion to clarify scope, time frame, target audience, or desired depth **Classify the query into a complexity tier:** | Tier | Signals | Agents | Action | |------|---------|--------|--------| | Simple | Single factua