AppSecGate MCP

Local-first security scanner for AI coding agents

✦ Offline · No code leaves your machine

When to use AppSecGate

✅ Use when

You’re using AI coding agents (Claude Code, Cursor, Cline, Continue) that generate or modify code on your machine
You want security scanning to happen before commit, not after
The project is sensitive and code must never leave your machine (offline-first)
You need an auditable trail of findings shown to the developer and their fix attempts
You want deterministic deduplication across scans — same finding, same fingerprint, no noise

❌ NOT when

You need a CI/CD gate that blocks builds — that’s Phase 2 (coming)
You want the server to auto-fix findings — AppSecGate finds, the LLM fixes
You need a cloud dashboard with team management — that’s Phase 3
Your project has zero code (docs-only, design files) — there’s nothing to scan

14 MCP tools

Tool	What it does
`scan_file`	Scan a single file and return normalized findings.
`scan_diff`	Scan staged + unstaged + untracked changes in the current git repository.
`scan_project`	Scan the entire project or a specific directory inside it.
`get_fix_context`	Return the ContextPack for a finding (finding + code window + rule doc).
`get_fix_context_batch`	Return ContextPacks for multiple findings (limit 20).
`verify_fix`	Verify a single finding: rescan the file and match by logical ID.
`verify_fix_batch`	Verify up to 20 findings in one operation; each file is rescanned once.
`query_findings_db`	Read-only SELECT on public views (v_findings, v_summary, v_projects).
`export_snapshot`	Export a snapshot of findings in ndjson format (parquet in Phase 2).
`get_config` / `update_config`	Read and update the active session configuration.
`submit_feedback`	Submit feedback on a finding: FALSE_POSITIVE / CONFIRMED / WONT_FIX.
`get_scan_history`	Return the shown findings log with verification attempts.
`suppress_finding`	Create a local suppression (does not affect CI gate in Phase 1).

Quick Install

Claude Code

Add to ~/.claude/claude_dotfiles/claude.json:

{
  "mcpServers": {
    "appsecgate": {
      "command": "docker",
      "args": ["run","--rm","-i","--network=none",
        "--read-only","--tmpfs=/tmp:size=100M,mode=1777",
        "--cap-drop=ALL",
        "--security-opt=no-new-privileges:true",
        "-v","/path/to/project:/workspace:ro",
        "-v","appsecgate-cache:/home/appsecgate/.cache",
        "-v","appsecgate-data:/home/appsecgate/.appsecgate-mcp",
        "moondoom/appsecgate-mcp:0.1.0"],
      "env": {}
    }
  }
}

Cursor

.cursor/mcp.json:

{
  "mcpServers": {
    "appsecgate": {
      "type": "docker",
      "command": "moondoom/appsecgate-mcp:0.1.0",
      "args": ["--transport","stdio","--project","/workspace"],
      "volumes": [
        {"host":"/path/to/project",
         "container":"/workspace","mode":"ro"},
        {"host":"appsecgate-cache",
         "container":"/home/appsecgate/.cache"},
        {"host":"appsecgate-data",
         "container":"/home/appsecgate/.appsecgate-mcp"}
      ]
    }
  }
}

Cline

.cline/mcp.json:

{
  "mcpServers": {
    "appsecgate": {
      "command": "docker",
      "args": ["run","--rm","-i","--network=none",
        "--read-only","--tmpfs=/tmp:size=100M,mode=1777",
        "--cap-drop=ALL",
        "--security-opt=no-new-privileges:true",
        "-v","/path/to/project:/workspace:ro",
        "-v","appsecgate-cache:/home/appsecgate/.cache",
        "-v","appsecgate-data:/home/appsecgate/.appsecgate-mcp",
        "moondoom/appsecgate-mcp:0.1.0"],
      "env": {}
    }
  }
}

Continue

.continue/mcpServers/appsecgate.yaml:

mcpServers:
  appsecgate:
    command: docker
    args:
      - run
      - --rm
      - -i
      - --network=none
      - --read-only
      - --tmpfs=/tmp:size=100M,mode=1777
      - --cap-drop=ALL
      - --security-opt=no-new-privileges:true
      - -v
      - /path/to/project:/workspace:ro
      - -v
      - appsecgate-cache:/home/appsecgate/.cache
      - -v
      - appsecgate-data:/home/appsecgate/.appsecgate-mcp
      - moondoom/appsecgate-mcp:0.1.0

How it works

Scan

Agent calls scan_file, scan_diff, or scan_project. Semgrep (SAST), Gitleaks/TruffleHog (secrets), Trivy (SCA) run in isolated subprocesses.

Fingerprint

Each finding gets a deterministic logical ID. Repeated scans never create duplicates. Suppressions and escalations stay attached across the finding's lifecycle.

Context

Agent requests get_fix_context — receives the code window, rule docs, and raw finding. No LLM calls on the server — just structured data.

Verify

After fixing, verify_fix rescans and matches by logical ID. Results logged in shown_findings.db for audit.