@actalumen/mcp-server
OfficialClick on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@@actalumen/mcp-server verify my contract for SOC2 compliance"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
@actalumen/mcp-server
MCP server for ActaLumen — verify documents for compliance from Claude Desktop, Cursor, and Claude Code.
What it does
Exposes ActaLumen's verification, chat, and document tooling to any MCP-compatible AI agent. Your agent can upload a contract, run it against a compliance template (SOC2, GDPR, custom), and chat with the document — all grounded in citations.
PII is redacted server-side before storage. Agents never see un-redacted content, even when explicitly asked.
Related MCP server: PII Redaction MCP Server
Install
Easiest: MCPB bundle (Claude Desktop one-click)
Download actalumen-<version>.mcpb from the latest release and double-click it. Claude Desktop will prompt for your API key and upload directory, then install. No Node, no JSON editing.
Or via npm (Cursor, Claude Code, custom MCP clients)
Prerequisites
An ActaLumen API key with
read,write, andverifyscopes. Create one at app.actalumen.com → Settings → API Keys.Node.js 18.17 or newer.
A directory the agent is allowed to upload from. Default:
~/actalumen-inbox(create it:mkdir ~/actalumen-inbox).
Claude Desktop
Edit ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows):
{
"mcpServers": {
"actalumen": {
"command": "npx",
"args": ["-y", "@actalumen/mcp-server"],
"env": {
"ACTALUMEN_API_KEY": "ak_live_..."
}
}
}
}Restart Claude Desktop. You should see the ActaLumen tools in the tools menu.
Cursor
Add to ~/.cursor/mcp.json:
{
"mcpServers": {
"actalumen": {
"command": "npx",
"args": ["-y", "@actalumen/mcp-server"],
"env": { "ACTALUMEN_API_KEY": "ak_live_..." }
}
}
}Claude Code
claude mcp add actalumen -- npx -y @actalumen/mcp-server
# then set the key in your shell or settings:
export ACTALUMEN_API_KEY=ak_live_...Configuration
Env var | Default | Purpose |
| (required) | Org-scoped API key. |
|
| Override for self-hosted or staging. |
|
| Path allowlist — agents can only upload files from inside this directory. |
Tools
Tool | Purpose |
| Upload a PDF (must be inside |
| Poll a document's status — wait for |
| Find existing documents by name without re-uploading. |
| Discover available compliance templates (e.g., SOC2 Vendor). |
| Run a document against a template. Returns a |
| Fetch report status and findings. |
| Ask grounded questions about one or more documents. |
| Check plan quota before batch operations. |
Example agent flow
"Check whether
~/actalumen-inbox/acme-msa.pdfmeets our SOC2 vendor requirements and summarize the gaps."
The agent will:
list_templates→ find the SOC2 template.upload_document→ get a document ID, wait forREADYviaget_document.start_verification→ get ajobId.get_verification(polling) → receive the full report.Summarize the failed criteria back to you with page citations.
Verify your install
ACTALUMEN_API_KEY=ak_live_... npx -y @actalumen/mcp-server --health
# → ok — https://api.actalumen.com — org Acme Inc — upload dir /Users/.../actalumen-inboxLicense
MIT
Available Tools
8 toolschat_with_documentA
Ask a grounded question about one or more documents. Returns an answer with citations into the document. PII in source documents has already been redacted server-side, so answers cannot leak redacted data.
Inputs: documentIds (required), message (required), sessionId.
get_documentA
Get a document's status and metadata. Use this to poll after upload — call start_verification or chat_with_document only when status is READY. Status values: PROCESSING (embeddings running), READY (queryable), FAILED.
Inputs: documentId (required).
get_usageA
Check the organization's current plan, quota, and remaining allowance for documents, verifications, chat messages, and generated media. Call before kicking off expensive batch operations.
Inputs: no named parameters.
get_verificationA
Fetch verification job status and (when complete) the full report with per-criterion findings, citations, and pass/fail summary. Poll roughly every 5–10s after start_verification.
Inputs: jobId (required).
list_documentsA
List documents in the organization. Use this to find an existing document ID by name instead of re-uploading.
Inputs: limit, offset.
list_templatesA
List compliance criteria templates available to the organization (e.g., 'SOC2 Vendor', 'GDPR DPA'). Call this before start_verification to find the right templateId. Each template encodes a checklist of criteria a document is verified against.
Inputs: no named parameters.
start_verificationA
Run compliance verification on a document against a template (or ad-hoc criteria). Returns a jobId; the job runs async — poll get_verification until status is 'completed'. Requires the API key to have the 'verify' scope.
Inputs: documentId (required), templateId, criteria.
upload_documentA
Upload a PDF to ActaLumen for compliance verification. The server applies organization-configured PII redaction before storage — the agent will only ever see redacted content in subsequent calls. Returns a document ID; the document is PROCESSING until embeddings finish (poll get_document for status=READY).
Inputs: path (required), id.
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 8 tools
Each tool has a clearly distinct purpose. Upload, list, get status, verify, chat, check usage, list templates, and get verification results are all unique operations with no overlap in functionality.
All tool names follow a consistent verb_noun pattern (e.g., upload_document, get_document, list_documents, start_verification). No mixing of conventions or inconsistent verb styles.
8 tools is well-scoped for a document compliance verification service. Each tool earns its place, covering the essential operations without unnecessary bloat or missing critical steps.
The tool set covers the full lifecycle: upload, list, get status, verify, poll verification, chat with document, check usage, and list templates. Minor gaps like a delete or cancel operation exist, but the core workflow is complete.
Maintenance
Related MCP Connectors
Hosted document tools for agents: extract, parse, classify, convert, OCR, sign, and redact.
Deterministic runtime safety for AI agents: scan PII, gate tool actions, verify LLM output.
Document hosting and encrypted agent memory with multi-tenant persistence.
Compliance frameworks (SOC 2, ISO 27001, CMMC, NIST, more) delivered to AI agents as MCP tools.
Related MCP Servers
- AlicenseAqualityDmaintenanceCompliance and guardrails infrastructure for AI agents, enabling safe operations within regulatory boundaries like GDPR and EU AI Act.641 PyPIMIT
- FlicenseNot gradedqualityDmaintenanceEnables AI agents to redact PII from text, summarize redacted content, and manage custom redaction patterns across multiple languages.-
- AlicenseNot gradedqualityDmaintenanceEnables AI agents to scan, redact, and govern sensitive data before sending it to external tools through DLP, secure chat, and shadow AI discovery tools.AGPL 3.0
- FlicenseNot gradedqualityDmaintenanceRedacts PII from LLM traces and tool payloads before they leave review, enabling PII scanning, payload redaction, sensitive field classification, privacy receipts, and trace audit exports.-