Files
crewAI/lib/crewai/tests
Devin AI 909ebd869b feat: add MCP security layer for message signing, tool integrity, and replay protection
Addresses issue #4875 - MCP tool calling has no per-message authentication
or integrity verification.

Adds:
- MCPSecurityConfig: Pydantic model for security settings (agent identity,
  key pairs, Trust Authority keys, signing/verification toggles)
- MCPSecurityManager: Stateful manager handling ECDSA P-256 key generation,
  agent passport creation/signing, message signing/verification, tool
  integrity checks, and nonce-based replay protection
- Integration into MCPClient, MCPToolResolver, and all config models
  (MCPServerStdio, MCPServerHTTP, MCPServerSSE)
- 30 comprehensive tests covering all security features and graceful
  degradation when mcp-secure is not installed

Uses the mcp-secure library (optional dependency) which implements the
IETF draft-sharif-mcps-secure-mcp specification.

Co-Authored-By: João <joao@crewai.com>
2026-03-14 17:47:51 +00:00
..
2026-03-04 00:45:09 -05:00
2025-10-20 14:10:19 -07:00
2025-12-04 16:53:19 -05:00
2026-02-13 21:34:37 -03:00
2025-10-20 14:10:19 -07:00
2026-02-13 21:34:37 -03:00
2025-12-04 13:34:29 -08:00
2025-10-20 14:10:19 -07:00
2025-12-01 18:56:56 -05:00
2025-10-20 14:10:19 -07:00
2025-10-20 14:10:19 -07:00
2025-10-20 14:10:19 -07:00