MCP Server · Now Available

Your AI agent,
meets software intelligence.

Connect any MCP-compatible AI assistant to CAST Highlight. Ask CAST about software health, cloud maturity, open source risks, and AI acceleration insights of your application portfolio — conversationally.

🤖 Ask CAST now! View Configurations
Works with
Anthropic Claude ✦ Google Gemini Any MCP Client
30+
meta tools exposed
REST
CAST Highlight API-native
0
Extra infrastructure
highlight-mcp — claude
# Claude asking the MCP server
Claude → get_portfolio_summary
 
{
  "applications": 247,
  "avg_health_score": 72.4,
  "cloud_maturity": 89,
  "critical_oss_risk": 12,
  "ai_ready_apps": 34
}
 
Claude → get_app_highlights
           appId="billing-core"
 
✓ Agentic Readiness: High
Cloud Maturity score: 84 · OSS Safety: Low
Tech debt: 3.2k violations ·
Why MCP

Portfolio intelligence.,
in every AI conversation.

Stop switching contexts. Your AI assistant can now query CAST Highlight directly — surfacing the signals that matter, when they matter.

🤖
Conversational portfolio queries
Ask your AI assistant "which apps have the highest cloud readiness scores?" or "show me open-source risk outliers" — and get structured answers backed by live Highlight data.
⚡
AI Acceleration Insights
Surface AI Opportunity scores and Agentic Readiness ratings directly from your AI agent. Plan your gen-AI investments with data, not guesswork.
🔒
Secure, token-based auth
Authenticate once via your Highlight API token. No credentials are stored in the MCP server — your data stays in your control.
📊
Cross-domain analysis
Combine application health, software composition, technology currency, and cloud readiness signals in a single agent workflow. The MCP server bridges all of Highlight's analytics domains into one coherent interface for your AI.
# Example agent query
user: "Which Java apps are
      candidates for agentic
      refactoring this sprint?"

→ highlight-mcp calls:
  get_applications_by_tech
  get_agentic_readiness
  get_cloud_maturity_score
🧩
MCP standard compliant
Works with any client implementing the Model Context Protocol — Claude Desktop, Cursor, Continue, and more. One install, every AI tool.



Prompts you can try.

Are any applications still using the log4j component? Drill into Adempiere. Give me a 30-second executive diagnosis: what are its three biggest risks, what evidence supports them? Looking at our portfolio in Highlight, which applications should we consider moving to cloud first and why? Which applications are likely costing me most to operate on the cloud? Which applications are the best candidates for retirement or replacement?
Quick Start

SaaS — Up and running in under a minute.

Three steps. No infrastructure. Just your CAST Highlight credentials and an MCP-compatible client.

1
Open your preferred AI client
Ask CAST supports most of AI assistants (Anthropic Claude, Google Gemini, OpenaI Codex, etc.)
2
Add and configure the MCP server
The three parameters below will be required to configure Ask CAST.
HIGHLIGHT_API_KEY=xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx HIGHLIGHT_DOMAIN=123456 MCP_URL=https://demo.casthighlight.com/mcp
3
Run your AI assistant and start asking CAST
YOU'RE ALL SET UP!
Claude Code mcp.json JSON
{
  "mcpServers": {
    "AskCAST": {
      "url": "https://demo.casthighlight.com/mcp",
      "headers": {
        "highlight_domain":
          "<your_domain_id>",
        "highlight_api_key":
          "<your_cast_highlight_user_token>"
      }
    }
  }
}
✓ You're ready to go
Ask your AI: "Summarize the software health of my portfolio" — and CAST Highlight will answer through your agent.
Client Setup

Ready-to-go configs
for every AI assistant.

Copy the configuration for your tool, drop it in the right file, and you're connected. No guesswork on syntax or file paths.

Claude Desktop
Recommended Anthropic's desktop application. Supports remote MCP servers natively via a JSON config file — no additional plugins needed.
Config file location
Windows: %APPDATA%\Claude\claude_desktop_config.json
claude_desktop_config.json 
"mcpServers": {
  "AskCAST": {
    "command": "npx",
    "args": [
      "mcp-remote@latest",
      "[SERVER_URL]/mcp",
      "--header",
      "highlight_api_key:${HIGHLIGHT_API_KEY}",
      "--header",
      "highlight_domain:${HIGHLIGHT_DOMAIN}"
    ],
    "env": {
      "HIGHLIGHT_API_KEY": "[USER_TOKEN]",
      "HIGHLIGHT_DOMAIN": "[DOMAIN_ID]"
    }
  }
}
Claude Code
Anthropic's agentic CLI for terminal-based development workflows. MCP servers are registered via the claude mcp add command or a project-level config file.
Config file location
Project root: .claude/.mcp.json
.claude/.mcp.json 
{
  "mcpServers": {
    "AskCAST": {
      "type": "http",
      "url": "[SERVER_URL]/mcp",
      "headers": {
        "highlight_domain": "[DOMAIN_ID]",
        "highlight_api_key": "[USER_TOKEN]"
      }
    }
  }
}
Cursor
The AI-first code editor. MCP servers are configured in Cursor's settings UI or in a JSON file under your project or global config directory.
Config file location
Global: ~/.cursor/mcp.json · Project: .cursor/mcp.json
.cursor/mcp.json 
{
  "mcpServers": {
    "AskCAST": {
      "url": "https://rpa.casthighlight.com/mcp/sse",
      "headers": {
        "Authorization": "Bearer hl_live_7f3a91bc2e084d6f_Kx9mTqVpNzR2wYsL4uJhDcBgAeF8iO"
      }
    }
  }
}
VS Code (GitHub Copilot)
MCP support is available in VS Code via the GitHub Copilot extension (Agent mode). Add the server to your VS Code user settings or workspace settings.
Config file location
.vscode/mcp.json
.vscode/mcp.json 
{
  "inputs": [
    {
      "type": "promptString",
      "id": "highlight-domain",
      "description": "CAST Highlight domain/company ID"
    },
    {
      "type": "promptString",
      "id": "highlight-api-token",
      "description": "CAST Highlight user token",
      "password": true
    }
  ],
  "servers": {
    "AskCAST": {
      "type": "http",
      "url": "[SERVER_URL]/mcp",
      "headers": {
        "Accept": "application/json, text/event-stream",
        "highlight_domain": "${input:highlight-domain}",
        "highlight_api_key": "${input:highlight-api-token}"
      }
    }
  }
}
OpenAI Codex CLI
OpenAI's terminal-based agentic coding assistant. MCP servers are declared in a config.yaml file in the Codex config directory.
Config file location
~/.codex/config.toml
~/.codex/config.toml 
[mcp_servers."AskCAST"]
url = "[SERVER_URL]/mcp"
trust = true
startup_timeout_sec = 30
tool_timeout_sec = 300

http_headers = {
  Accept = "application/json, */*;q=0.1",
  highlight_domain = "[DOMAIN_ID]",
  highlight_api_key = "[USER_TOKEN]"
}
Google Gemini CLI
Google's Gemini CLI supports MCP servers via a settings.json file. Once registered, Gemini can call CAST Highlight tools as part of any agentic task.
Config file location
~/.gemini/settings.json
~/.gemini/settings.json 
"mcpServers": {
    "AskCAST": {
      "httpUrl": "[SERVER_URL]/mcp",
      "headers": {
        "Accept": "application/json, */*;q=0.1",
        "highlight_api_key": "[USER_TOKEN]",
        "highlight_domain": "[DOMAIN_ID]"
      },
      "trust": true,
      "timeout": 30000
    }
}
Replace the token, domain, and MCP server url placeholders.
The token [USER_TOKEN], domain [DOMAIN_ID], and MCP server url [SERVER_URL] shown in configs above is for illustration only. Retrieve your real domain and API token from CAST Highlight → User Account → Generate Access Token and substitute it before connecting.