Deep Research Team

    4

    Deploy a hierarchical team of AI agents to perform 15-30 minute deep-dive research with parallel execution.

    Free

    245 uses4.8 (5 reviews)

    Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+21 more

    deep-research-team

    Example session with this skill installed

    Run a quick research report on the state of the AI agent ecosystem in 2025 across platforms, technology, and business models. Use Quick Mode for a high-level overview.

    • Read your context and instructions
    • Compiled the deep-research-team

    Quick Mode output structure
    ai_agent_ecosystem_2025/
    ├── 00_planning/research_plan.md
    ├── 01_platforms/research.md + sources.md (27 sources)
    ├── 02_technology/research.md + sources.md (30 sources)
    ├── 03_business/research.md + sources.md (25 sources)
    └── final_report/executive_summary.md

    Each research.md contains structured findings with data tables, key conclusions, and cross-references. Each sources.md logs every cited source with URL, type, confidence rating, and citation location. The executive summary synthesizes all modules into a single overview with key metrics, core findings, cross-module consensus/contradictions, and action recommendations.

    sources.md

    Connects securely to your tools. The creator never sees your data.

    About this skill

    Advanced Multi-Agent Research Orchestration

    The Deep Research Team skill transforms your AI environment into a high-powered research laboratory. Instead of relying on a single prompt that may hallucinate or miss nuances, this skill deploys a structured team of specialized agents that work in parallel to investigate, cross-validate, and synthesize complex information into publication-ready reports.

    What It Does

    The skill automates a multi-phase research workflow designed for depth and accuracy:

    • Phase 0 (Planning): A PM Agent designs a custom research architecture and directory structure based on your specific topic.
    • Phase A (Parallel Research): Multiple specialized agents are dispatched simultaneously to investigate different facets of the topic, ensuring high throughput and diverse perspectives.
    • Phase B & C (Synthesis & Audit): A Synthesis Agent identifies patterns and contradictions across findings, while a Review Agent (powered by high-reasoning models) performs a critical quality audit.
    • Phase D (Delivery): A Final Report Agent assembles a comprehensive document with a separate executive summary and full source citations.

    Why Use This Skill?

    Standard AI prompting often lacks the breadth needed for professional due diligence. This skill is superior because it uses a file-system relay for agent communication, preventing the "context drift" associated with long chat threads. It supports both a Quick Mode for rapid overviews and a Full Mode for academic-grade investigations, complete with automated quality scoring and error-checking.

    Output & Capabilities

    The result is a structured workspace containing individual module findings, a synthesis report, and a final executive summary. Every fact is mapped to its primary source, making it ideal for competitive analysis, investment research, and technical due diligence.

    Full Mode (6-13 agents) consumes significant API usage. Quick Mode (4 agents) is recommended for initial exploration. Research quality depends on web search availability. Reports default to English but support any language.

    How to use

    1. 1

      Connect your agent to the Agensi MCP

      Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup

    2. 2

      Sign in to Agensi

      Your agent shows a sign-in link. Sign in and click Allow.

    3. 3

      Ask for what you need

      Your agent loads this skill and uses it on your task. It always asks before any payment.

    Reviews

    4.8
    5 reviews
    5
    4
    4
    1
    3
    0
    2
    0
    1
    0

    Trust & safety

    Security scan

    Security scan passed 7 months ago

    • Free with an Agensi account

    Permissions

    Read Files
    Write Files
    Network Access

    File Scopes

    *.md

    Read Files: reads prompt templates and research outputs. Write Files: creates research reports and source files. Network Access: research agents use web search to gather sources.

    Listed7 months ago

    What's inside

    Frequently Asked Questions

    Unlike standard AI chats that work linearly, this skill uses a hierarchical team of agents that research different sub-topics in parallel. This prevents 'context drift' and ensures a much higher level of detail, cross-validation, and source-checked accuracy than a single LLM could provide alone.

    The skill is designed to work with high-reasoning models (like GPT-4o or Claude 3.5 Sonnet) that support tool-use and file-system interactions. It functions as an orchestration layer that can be integrated into most major AI agent frameworks that allow for multi-agent workflows.

    You'll receive the complete orchestration logic for the PM, Synthesis, Review, and Reporting agents, along with the file-system relay configuration. The package also includes pre-configured templates for both 'Quick Mode' and 'Full Mode' research sessions.

    While a single agent might take 5-10 minutes for a broad topic, the parallel execution of this team typically delivers a comprehensive, multi-chapter report in 15-30 minutes. The depth of the 'Full Mode' is limited only by your model's rate limits and the complexity of the research architecture designed by the PM Agent.

    This skill is best suited for complex, data-heavy tasks like market intelligence, competitive analysis, investment due diligence, and technical landscape mapping. It is designed to solve the problem of surface-level AI 'hallucinations' by requiring cross-agent verification and source-mapped reporting.

    New in Agents & Orchestration

    Free