MCP Integration¶
Salient is built for the age of AI agents. It exposes a full MCP (Model Context Protocol) server with 40+ tools that let any compatible AI agent manage exercises, query the digital twin, sync connectors, and compile security intelligence.
Salient is also an MCP client — it can consume data from external MCP servers (Gmail, Notion, Google Calendar, Slack, etc.) and feed it into your digital twin.
Quick Setup¶
One command:
Or add to your project .mcp.json:
Add to your Claude Desktop config (~/Library/Application Support/Claude/claude_desktop_config.json):
Salient's MCP server uses stdio transport and follows the standard MCP protocol. Any client that supports stdio MCP servers can connect:
- Command:
python3 tools/mcp-ttx/server.py - Environment:
SALIENT_API_URL,SALIENT_TOKEN - Protocol: JSON-RPC 2.0 over stdin/stdout
All Platforms¶
| Platform | Transport | Guide |
|---|---|---|
| Claude Code | stdio | Setup |
| Claude Desktop | stdio | Setup |
| ChatGPT / OpenAI | HTTP | Setup |
| Google Gemini | stdio / HTTP | Setup |
| Amazon Bedrock | HTTP | Setup |
| Cursor | stdio | Setup |
| VS Code / Copilot | stdio | Setup |
| Windsurf | stdio / HTTP | Setup |
| JetBrains IDEs | stdio | Setup |
| Cline | stdio | Setup |
| Remote / Hosted | HTTP | Setup |
What You Get¶
Once connected, your AI agent has access to:
| Category | Tools | What They Do |
|---|---|---|
| Exercises | list_scenarios, get_scenario, create_scenario, save_session, save_evaluation | Browse, run, and score tabletop exercises |
| Twin Intelligence | query_twin, twin_facts, twin_facts_summary, detect_patterns, recommend_scenario | Query the digital twin, detect recurring gaps, get scenario recommendations |
| Posture | posture_assessment, posture_report, posture_timeline | Generate board-ready security posture assessments |
| Connectors | okta_sync, entra_id_sync, list_connectors, twin_coverage, discover_connector | Sync identity data, analyze coverage gaps |
| Threat Intel | ingest_threat_intel, list_threat_intel, generate_scenario_from_intel | Ingest threat articles, generate targeted scenarios |
| MCP Ingestion | ingest_from_mcp, ingest_document, list_mcp_sources | Feed data from sibling MCPs into the twin |
| Playbooks | generate_playbook, get_playbook | Generate incident response playbooks from exercise results |
| Dashboard | get_dashboard, get_org_profile | Get platform stats and org profile |
The /ttx Skill¶
If you're using Claude Code, the /ttx skill provides guided workflows:
/ttx Run a tabletop exercise
/ttx profile Review digital twin completeness
/ttx gather Guided evidence collection
/ttx evaluate Re-evaluate a past exercise
/ttx playbook Generate an incident response playbook
/ttx enrich Pull data from available sibling MCPs
/ttx status Dashboard overview
Two-Way MCP¶
Salient doesn't just serve data to AI agents — it consumes data from them.
Path 1: AI-Mediated — When Claude has access to Gmail, Calendar, Notion alongside Salient, the /ttx enrich command orchestrates across all of them. Security-relevant emails, policy documents, calendar events flow into your twin automatically.
Path 2: Direct Client — Salient's backend connects to external MCP servers directly, discovers their tools and resources, and syncs data into the twin.