tridentchain-mcp

tridentchain-mcp

MCP server for TridentChain Security — one local stdio server for Claude, Cursor, VS Code, Windsurf, Zed, and any MCP-compatible agent.

Requires: pip install "tridentchain-security>=0.1.1" (unified tool layer).

The tridentchain-security CLI is unchanged — universal fallback for any agent that runs shell commands.

Install

Option A — pip (Python 3.10+)

pip install -U "tridentchain-security>=0.1.4" "tridentchain-mcp>=0.1.4"

Option B — uvx (no manual Python install)

uvx tridentchain-mcp

Option C — Docker (no Python needed on host)

docker pull ghcr.io/devinder1/tridentchain-mcp:latest
docker run --rm -i -v "$PWD:/workspace" ghcr.io/devinder1/tridentchain-mcp:latest

Public image: https://github.com/DevInder1/supply-chain-scanner-public/pkgs/container/tridentchain-mcp — linux/amd64 and linux/arm64, MIT, non-root user, ~205 MB.

Tools (all agents)

Tool	Description
scan_project	Project dependencies only
scan_full	Project + system + extensions
validate_after_patch	Diff two scan JSON results

Quick config

MCP hosts using mcpServers (Claude, Cursor, Windsurf):

{
  "mcpServers": {
    "tridentchain": {
      "command": "python3",
      "args": ["-m", "tridentchain_mcp"]
    }
  }
}

Or with Docker (no Python required on host):

{
  "mcpServers": {
    "tridentchain": {
      "command": "docker",
      "args": [
        "run", "--rm", "-i",
        "-v", "${workspaceFolder}:/workspace",
        "ghcr.io/devinder1/tridentchain-mcp:latest"
      ]
    }
  }
}

VS Code uses "servers" — see integrations/vscode/mcp.json.example.

OpenAI uses function calling (not MCP) — see integrations/openai/ and examples/openai/.

Agent setup

Agent	Guide
All platforms	docs/AGENT_INTEGRATIONS.md
Claude	docs/CLAUDE_MCP_SETUP.md
Cursor	docs/CURSOR_SETUP.md
VS Code	docs/VSCODE_SETUP.md
Helper script	./scripts/setup-agent-mcp.sh <agent>

Copy-paste configs: integrations/mcp/