Add Salient to ChatGPT¶
Connect Salient's MCP server to ChatGPT for AI-powered tabletop exercise management directly in OpenAI's interface.
Remote MCP server required
ChatGPT does not support stdio transport. You must host Salient's MCP server as an HTTP endpoint using Streamable HTTP or SSE transport. See the Remote MCP Server guide for setup instructions.
Requirements¶
- ChatGPT Pro, Plus, Business, or Enterprise plan
- A Salient MCP server exposed over HTTPS (see Remote Setup)
Step 1: Enable Developer Mode¶
- Open ChatGPT Settings
- Navigate to Workspace Settings > Permissions
- Enable Developer Mode
Step 2: Create an MCP App¶
- Go to ChatGPT Apps settings (or visit
chatgpt.com/gpts/mine) - Click Create App
- Select MCP Server as the app type
- Enter your Salient MCP server URL:
- Add authentication headers if your remote server requires them
- Save and enable the app
Step 3: Use in Chat¶
Once the app is enabled, start a new conversation and mention Salient tools:
"List my tabletop scenarios" "Query the digital twin for our MFA coverage" "Run a posture assessment"
ChatGPT will invoke Salient tools through the MCP connection.
Programmatic Access via Responses API¶
For automation, use OpenAI's Responses API with MCP tools:
from openai import OpenAI
client = OpenAI()
response = client.responses.create(
model="gpt-4.1",
tools=[{
"type": "mcp",
"server_label": "salient",
"server_url": "https://your-salient-host/mcp",
"headers": {
"Authorization": "Bearer your-salient-token"
}
}],
input="List all tabletop exercise scenarios"
)
print(response.output_text)
Transport Notes¶
ChatGPT supports two MCP transport modes:
| Transport | Status |
|---|---|
| Streamable HTTP | Supported (recommended) |
| SSE | Supported (legacy) |
| stdio | Not supported |
Hosted endpoint coming soon
Salient will offer a hosted MCP endpoint for paid plans, eliminating the need to self-host the HTTP server.