Quick Start¶
Get Salient running and complete your first tabletop exercise in 5 minutes.
Prerequisites¶
- Docker and Docker Compose
- Python 3.11+ (for MCP server)
- A Claude API key (optional — platform works without it, AI features need it)
1. Clone and Start¶
git clone https://github.com/noperse/projectTTX_Salient.git
cd projectTTX_Salient
cp .env.example .env
Edit .env with your settings:
# Required
SECRET_KEY=$(python3 -c "import secrets; print(secrets.token_hex(32))")
# Optional — enables AI scoring, playbook generation, and fact mining
AI_PROVIDER=anthropic
ANTHROPIC_API_KEY=sk-ant-...
Start the platform:
This starts the backend (:8000) and frontend (:3000).
2. Connect MCP¶
In your project directory, the .mcp.json is already configured. If using Claude Code:
You should see ttx-salient in the list.
3. Run Your First Exercise¶
Claude will:
- Load the evaluator framework
- Pull your org profile (empty on first run — that's fine)
- Show available scenarios
- Facilitate the exercise inject-by-inject
- Score your responses against NIST CSF
- Mine organizational facts from your answers
- Generate an incident response playbook
4. Build Your Twin¶
After your first exercise, the twin starts learning. Accelerate it:
This walks you through collecting evidence: security tools, team contacts, IR plans, backup procedures. Every piece enriches the twin.
5. Connect Identity Data¶
If you have Okta or Entra ID:
The connector setup wizard walks you through it in the browser, or Claude can trigger syncs via MCP tools.
What's Next¶
- Run more exercises — each one makes the twin smarter and the scoring more targeted
- Upload artifacts — IR plans, security policies, architecture docs (drag-and-drop in the UI)
- Connect sibling MCPs —
/ttx enrichpulls security data from Gmail, Notion, Calendar - Review posture —
posture_assessmentgives you a board-ready security report