Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIDeepSeekDeepSeekMuseMuse+22 more

    Deep Research Protocol

    1

    Auditable, evidence-backed multi-agent deep research protocol with persistent claim-traceability and file-backed reports.

    $4.99

    Secure checkout via Stripe

    Money-back guarantee

    Converts to your local currency at checkout

    Security scanned
    Deep Research Protocol

    Deep Research Protocol

    Example session with this skill installed

    Run a deep research investigation comparing the architectural tradeoffs between dense retrieval and sparse/BM25 retrieval in production RAG systems. Verify claims with independent sources.

    • Read your context and instructions
    • Compiled the deep research protocol

    Summary

    The protocol orchestrated 6 independent research workstreams, audited 14 primary technical papers and production benchmarks, and produced a fully verifiable report.

    DimensionDense Retrieval (e.g. BGE/OpenAI)Sparse Retrieval (BM25 / SPLADE)Hybrid (RRF)
    Exact Keyword Recall68.2% [[C004]]89.4% [[C001]]93.1% [[C009]]
    Semantic & Synonym Match91.5% [[C003]]54.1% [[C002]]94.8% [[C008]]
    Index Latency (P99)14.2ms [[C005]]3.1ms [[C007]]18.5ms [[C010]]

    Evidence Status

    • Verified claims: 18 (100% backed by primary benchmarks)
    • Contradictions resolved: 2 (latency divergence across GPU vs CPU deployment)
    • Deliverable: Audit-passed report rendered to article/report.md with deterministic citation bibliography.

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

    About this skill

    Deep Research Protocol

    Run substantial research as a persistent, evidence-backed multi-agent project rather than a fragile single-prompt interaction.

    Designed specifically for researchers, analysts, academics, and power developers reading papers and conducting complex multi-source investigations. The unit of quality is a supported claim, not a polished paragraph.


    🌟 Why Deep Research Protocol?

    Generic "deep research" prompts suffer from context loss, unverified claims, and hallucinated citations. Deep Research Protocol solves this through strict multi-agent division of labor:

    • 🔍 Claim-Evidence Traceability: Every factual proposition carries an explicit claim marker (e.g. [[C017]]) backed by direct primary evidence or two independent sources.
    • 🛡 Zero-Hallucination Writer Boundary: The non-browsing Writer works only from the verified research packet—it is strictly prohibited from web-browsing or inventing facts.
    • 👥 Multi-Agent Orchestration:
      • Lead Agent: Plans, decomposes workstreams, and integrates findings.
      • Research Subagents: Work independently and save persistent source captures.
      • Writer: Drafts structured reports strictly bounded by the evidence graph.
      • Verifier: Independently audits the report against source policy gates.
    • 📁 Persistent File-Backed Workspace: Generates comprehensive local artifacts, structured handoffs, registries, and reproducible output.

    📦 What's Included

    • SKILL.md: Complete execution constitution and orchestration logic.
    • references/: 6 modular specifications (source-policy, subagent-contracts, writing-protocol, schemas, new-mode, audit-mode).
    • scripts/: Ready-to-run Python validation and markdown rendering utilities (validate_run.py, render_report.py).
    • tests/: Test suite ensuring protocol verification integrity.

    Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Research & Analysis.

    Requires host support for subagent delegation. Needs Python 3 to run validation and rendering scripts. Does not perform simple factual lookups or short web searches.

    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 buying a skill.

    Reviews

    No reviews yet

    Be one of the first to try it. Every listed skill goes through the trust checks below.

    Security scanned

    Checked by our 8-point scan before listing

    Fresh listing

    Recently published to Agensi

    Money-back guarantee

    Doesn't do what its page promises? You get your money back.

    Trust & safety

    Security scan

    Security scan passed 2 days ago

    Permissions

    Terminal / Shell
    Read Files
    Write Files
    Listed2 days ago

    What's inside

    Frequently Asked Questions

    Standard AI research often blends training data with real-time results; this protocol enforces a 'Strict Synthesis Boundary' that prevents the agent from using its own memory, ensuring every sentence is derived from captured evidence.

    This skill is compatible with orchestration frameworks that support multi-agent delegation and tool-use, such as LangGraph, CrewAI, or custom implementations using MCP servers.

    You will receive the complete multi-agent workflow logic, the structured schema for evidence registries, and the Python-based validation and rendering scripts.

    Yes, the protocol includes a 'Capability Preflight' step that allows you to configure specific MCP servers for search, browser automation, or local database access depending on your needs.

    The 'Independence Verification' feature clusters sources to ensure that multiple citations aren't just echoing the same primary source, providing a true measure of consensus or disagreement.

    Basic setup involves configuring your environment for Python 3 and connecting your preferred search or browsing tools; the included preflight script will help you verify your configuration before your first run.

    New in Research & Analysis