New

Diligence intelligence, inside your AI.

Connect Sigma Diligence to Claude, ChatGPT, or any LLM via the Model Context Protocol (MCP). Query findings, search documents, generate memos, and run agents - all from your AI assistant.

Generate API key mcp.sigmadiligence.com

How it works

Sigma exposes its diligence engine as a set of tools your AI can call. Authenticate once with an API key - your LLM does the rest.

Step 1

Connect

Add Sigma as an MCP server in your client (Claude Desktop, Cursor, custom agent) or call the REST API with your API key.

Step 2

Query

Your LLM discovers available tools and calls them when relevant - fetching findings, searching documents, reading memos.

Step 3

Act

Trigger actions: run the diligence agent, generate IC memos, detect contradictions - all orchestrated by your AI.

Simple configuration

Add Sigma to your MCP client config with your API key. Your LLM automatically discovers all available tools and knows when to call them.

MCP endpoint:mcp.sigmadiligence.com

API key authentication - no OAuth flows required

Read-only and write scopes for fine-grained access

Rate-limited and audit-logged for compliance

Works with any MCP-compatible client

claude_desktop_config.json

{
  "mcpServers": {
    "sigma-diligence": {
      "url": "https://mcp.sigmadiligence.com/mcp",
      "headers": {
        "Authorization": "Bearer sk_live_..."
      }
    }
  }
}

Available tools

16 tools covering the full diligence lifecycle - from raw documents to IC-ready memos.

Projects

list_projects

List all diligence projects for the authenticated firm with status, deal stage, and workstream summary.

filter?: { status, dealStage }

get_project

Fetch a single project with full workstream breakdown, decision readiness score, and hypothesis coverage.

projectId: string

Intelligence

get_findings

Retrieve all agent findings for a project or workstream, with risk ratings, evidence, and review status.

projectId, workstreamId?, riskRating?

get_contradictions

List open contradictions - conflicting claims across evidence sources with severity and hypothesis links.

projectId, status?

get_hypotheses

Fetch the hypothesis set with materiality, confidence scores, and evidence requirement status.

projectId, status?

get_decision_state

Current decision readiness: score, recommendation (advance/hold/stop), confidence by workstream, unresolved critical count.

projectId

Reports

get_factpack

Retrieve the Fact Pack & Growth Assessment - AI-researched company overview with market, competitive, and financial analysis.

projectId

get_memo

Fetch the latest workstream or stage memo (versioned), or list all memo versions for a workstream.

projectId, workstreamId?, version?

get_report_section

Retrieve a specific deep research section (executive summary, commercial, technology, etc.) in markdown.

projectId, sectionId

Data Room

search_documents

Semantic search across data room documents - returns relevant chunks with source attribution and confidence scores.

projectId, query, topK?

list_documents

List all files in the data room organized by workstream folder, with processing status and metadata.

projectId, workstreamType?

get_interviews

List completed interview sessions with transcripts, interviewee details, and extracted Q&A notes.

projectId, workstreamId?

get_question_framework

Retrieve the diligence question framework for a workstream, including assignments, responses, and evidence links.

projectId, workstreamId

Actions

generate_memo

Trigger IC-grade memo generation for a workstream or overall stage. Returns the versioned memo when complete.

projectId, workstreamId?

run_agent

Start the AI agent on a workstream - answers all framework questions from available evidence and generates findings.

projectId, workstreamId

detect_contradictions

Trigger cross-reference analysis to detect contradictions across all claims and evidence.

projectId

What you can do

Natural language questions backed by structured diligence data.

Ask questions about a live deal

"What are the top 3 risks in the BlueCat commercial diligence?"

Your LLM pulls findings, contradictions, and hypothesis data in real time - no copy-paste from reports.

Generate IC-ready analysis

"Write a one-page summary comparing retention across all portfolio companies."

The LLM fetches structured data from multiple projects and synthesizes cross-deal intelligence.

Research with full context

"Based on the data room, what does the customer churn look like for Q3?"

Semantic search across uploaded documents returns the exact paragraphs and tables the LLM needs.

Automate diligence workflows

"Run the agent on commercial, then generate the memo when it finishes."

Chain tool calls to orchestrate multi-step diligence: run agents, detect contradictions, produce memos.

Works with your stack

MCP is an open standard. Sigma works with any client that speaks the protocol, plus a REST API for everything else.

Claude (Anthropic)

Use Sigma as an MCP server in Claude Desktop, Claude Code, or any MCP-compatible client.

OpenAI / ChatGPT

Connect via function calling or the Assistants API with Sigma tools as external functions.

Custom LLM Agents

Any agent framework (LangChain, CrewAI, AutoGen) can call Sigma tools via the REST API or MCP protocol.

IDE & Dev Tools

Cursor, VS Code, Windsurf, and other MCP-enabled editors can query diligence data inline.

Authentication

Every tool call is authenticated with your firm’s API key. Keys are scoped to your firm’s data - you only see projects your team owns. Generate keys from your Sigma dashboard under Settings.

# REST API example

curl https://mcp.sigmadiligence.com/v1/projects \
  -H "Authorization: Bearer sk_live_..." \
  -H "Content-Type: application/json"

Base URL: mcp.sigmadiligence.com

Request Early Access

Frequently asked questions

What is MCP?

Model Context Protocol is an open standard created by Anthropic that lets AI assistants connect to external tools and data sources. Instead of copy-pasting data into prompts, your AI can query Sigma directly.

Do I need to change how I use Sigma?

No. MCP is an additional interface on top of Sigma. Your dashboard, reports, and workflows work the same way. MCP adds the ability for external AI tools to read and act on your diligence data.

Is my data safe?

Yes. API keys are scoped to your firm. All requests are authenticated, rate-limited, and audit-logged. Data never leaves your Sigma tenant - the AI queries your data in place.

Which AI tools support MCP?

Claude Desktop, Claude Code, Cursor, Windsurf, and any custom agent built with the MCP SDK. For tools that don't support MCP natively (like ChatGPT), use the REST API with function calling.

Can the AI modify my data?

Read-only and write scopes are separate. By default, API keys are read-only. Write access (run agents, generate memos) requires explicit opt-in when creating the key.

How do I get an API key?

Sign in, open Settings → API & MCP, and generate a key. It is shown only once - copy it into your MCP client (Authorization: Bearer) or store it in a secrets manager.