By defiect
Multi-agent deep research orchestrator with evidence graphs, quality gates, and publication-quality report generation.
npx claudepluginhub defiect/deep-research-pluginPerforms deep analysis for deep research runs. Reads sources thoroughly, extracts atomic claims with citations, builds evidence edges for triangulation, and documents conflicts between sources.
Orchestrates multi-agent Deep Research runs. Decomposes questions, assigns tasks to teammates, enforces evidence quality, performs adversarial review, and assembles final deliverables.
Performs wide-pass discovery for deep research runs. Searches the web using diverse queries, evaluates source candidates, captures metadata with credibility scoring, and writes structured notes.
Synthesizes deep research findings into publication-quality reports. Reads all claims, evidence, and notes to produce coherent, evidence-backed narratives with explicit citations, uncertainty, and conflict documentation.
Bootstrap the deep research workspace directory structure and verify the environment is configured correctly.
Re-render the deep research report from existing artifacts without re-fetching sources. Useful for updating the report format or regenerating after manual edits to claims/evidence.
Run an end-to-end deep research workflow. Multi-agent orchestration with evidence tracking, triangulation, quality gates, and a final cited report.
Check the progress of the current or most recent deep research run. Shows stage, source counts, claim counts, and any issues.
A Claude Code plugin for multi-agent deep research with evidence graphs, quality gates, and publication-quality report generation.
Turns Claude Code into a deep research system that goes beyond simple summarization:
claude --plugin-dir /path/to/deep-research-plugin
Or add to your Claude Code settings to load it automatically.
export CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS=1
/deep-research:run What are the implications of the EU AI Act for US technology companies?
This kicks off the full pipeline: planning → source discovery → deep reading → claim extraction → triangulation → synthesis → adversarial review → report generation.
/deep-research:status
/deep-research:report
/deep-research:init
| Agent | Model | Role |
|---|---|---|
dr-lead | Opus | Orchestrates the run, decomposes questions, assigns tasks, performs adversarial review |
dr-scout | Sonnet | Wide-pass discovery: searches, evaluates sources, writes preliminary notes |
dr-analyst | Opus | Deep reading, atomic claim extraction, triangulation, conflict identification |
dr-writer | Opus | Synthesis and publication-quality report generation |
Each research run produces a directory under .deep-research/runs/ containing:
run.json — Run metadata and configuration
plan.md — Research plan with strands and query strategy
queries.json — Generated search queries by category
sources.jsonl — Sources with metadata and credibility scoring
notes/ — Structured notes per source
claims.jsonl — Atomic claims with citations
evidence.jsonl — Triangulation edges (supports/contradicts)
conflicts.md — Documented disagreements between sources
synthesis.md — Narrative synthesis
report.md — Final structured report
audit.json — Quality audit results
Hooks enforce artifact requirements:
The plugin works with built-in WebSearch and WebFetch by default. For enhanced capabilities, configure MCP servers for:
MIT
Uses power tools
Uses Bash, Write, or Edit tools
Share bugs, ideas, or general feedback.
PhD-level research capabilities: literature review, multi-source investigation, critical analysis, hypothesis-driven exploration, quantitative/qualitative methods, and lateral thinking
Semantic search for Claude Code conversations. Remember past discussions, decisions, and patterns.
Comprehensive skill pack with 66 specialized skills for full-stack developers: 12 language experts (Python, TypeScript, Go, Rust, C++, Swift, Kotlin, C#, PHP, Java, SQL, JavaScript), 10 backend frameworks, 6 frontend/mobile, plus infrastructure, DevOps, security, and testing. Features progressive disclosure architecture for 50% faster loading.
Comprehensive PR review agents specializing in comments, tests, error handling, type design, code quality, and code simplification
Develop, test, build, and deploy Godot 4.x games with Claude Code. Includes GdUnit4 testing, web/desktop exports, CI/CD pipelines, and deployment to Vercel/GitHub Pages/itch.io.