Standards that enable AI agents to communicate with tools, data, and each other.
▶ 5 protocols indexed
MCP(Anthropic)
v2025-11-05JSON-RPC over stdio/HTTP+SSEACTIVE
Capabilities:
- Tool execution
- Resource access
- Prompt templates
- Sampling
- Roots
Implemented by: 5 frameworks, 5 tools
A2A(Google / Linux Foundation)
v0.3HTTP, SSE, JSON-RPCACTIVE
Capabilities:
- Agent discovery
- Task management
- Streaming updates
- Push notifications
- Artifact exchange
Implemented by: 2 frameworks, 0 tools
Agent Skills(Anthropic (open standard))
v1.0File-based (SKILL.md)ACTIVE
Capabilities:
- Skill definition
- Progressive disclosure
- Modular capabilities
- Cross-tool portability
Implemented by: 1 framework, 3 tools
AGNTCY(Linux Foundation)
v0.1SLIM protocolBETA
Capabilities:
- Agent discovery
- Agent identity
- Agent messaging
- Agent observability
Implemented by: 0 frameworks, 0 tools
Agent Protocol(AI Engineer Foundation)
v1.0HTTP REST APIACTIVE
Capabilities:
- Task management
- Step execution
- Artifact handling
Implemented by: 0 frameworks, 0 tools
total 5 protocols