feat: add pre-execution MCP trust verification hook (#2501)

[pranav-afk]

Add a configurable pre-execution trust verification hook for MCP tool calls so operators can block, warn, or allow remote MCP servers before any tool action runs.

• What kind of change does this PR introduce? (Bug fix, feature, docs update, ...)

  Feature — pluggable security middleware for MCP tool execution.

• Why was this change needed? (You can also link to an open issue here)

  DocsGPT currently has no framework-level gate before MCP tool execution. A misconfigured or malicious MCP server could run arbitrary operations without a pre-flight trust check. As third-party MCP servers become more common, this is an important security surface. Closes Add pre-execution trust verification hook for MCP tool calls #2501.

• Other information:

  • New module application/security/mcp_trust.py with:
    • TrustVerifier protocol (sync or async)
    • TrustContext / TrustResult / TrustVerdict (allowed | blocked | warned)
    • set_trust_verifier() for custom backends (allowlist API, SBOM, policy engine, etc.)
    • Built-in AllowlistTrustVerifier (URI/host allowlist + denylist)
    • enforce_mcp_trust() raises MCPTrustBlockedError when blocked
  • Wired into MCPTool.execute_action before any call_tool network call
  • Opt-in by default — existing workflows unchanged until configured
  • Settings:
    • MCP_TRUST_VERIFICATION_ENABLED (default false)
    • MCP_TRUST_ALLOWED_SERVERS / MCP_TRUST_BLOCKED_SERVERS (CSV)
    • MCP_TRUST_ON_WARN (proceed | block)
    • MCP_TRUST_FAIL_OPEN (default true — verifier errors do not hard-fail)
  • Unit tests in tests/security/test_mcp_trust.py (20 passed)
  • Backend-only change (no UI screenshots)

  Custom verifier example:

  from application.security.mcp_trust import (
      TrustContext, TrustResult, TrustVerdict, set_trust_verifier,
  )
  
  class MyVerifier:
      def verify(self, context: TrustContext) -> TrustResult:
          return TrustResult(verdict=TrustVerdict.ALLOWED, reason="ok")
  
  set_trust_verifier(MyVerifier())

Type of change

• [ ] Bug fix (non-breaking change which fixes an issue)
• [x] New feature (non-breaking change which adds functionality)
• [ ] Breaking change (fix or feature that would cause existing functionality to not work as expected)
• [ ] This change requires a documentation update

How has this been tested?

• [x] Unit tests
• [x] Manual tests

Test Configuration:
• Python: 3.11
• Suite: tests/security/test_mcp_trust.py (20 passed)

Checklist:

• [x] My code follows the style guidelines of this project
• [x] I have already covered my code with unit tests
• [x] I have made corresponding changes to the documentation (N/A — settings documented in settings.py)
• [x] I have added an explanation of my changes

Related issues

Closes #2501

[vercel]

@pranav-afk is attempting to deploy a commit to the Arc53 Team on Vercel.

A member of the Team first needs to authorize it.