# Agent frameworks and SDKs

> 6 agent frameworks & SDKs ranked by the Anchor benchmark. Leader OpenAI Agents SDK (AA). Libraries that run the agent loop. Tool calling, MCP, multi-agent hand-offs, durable state, human approval and tracing. Compared on what they do by default, including the telemetry they send.

- Canonical: https://www.anchorterminal.com/categories/frameworks
- Markdown: https://www.anchorterminal.com/categories/frameworks.md (~1,800 tokens)
- Slim: https://www.anchorterminal.com/categories/frameworks.min.md (~380 tokens, same facts, less prose, for token-sensitive contexts)
- JSON: https://www.anchorterminal.com/categories/frameworks.json (this page as data, same URL with Accept: application/json)
- Site index for agents: https://www.anchorterminal.com/llms.txt (full text: https://www.anchorterminal.com/llms-full.txt)
- API: https://www.anchorterminal.com/api/v1/index.json
- Updated: 2026-10-04

Libraries that run the agent loop. Tool calling, MCP, multi-agent hand-offs, durable state, human approval and tracing. Compared on what they do by default, including the telemetry they send.

- Tools ranked: 6 · agent-ready (BB or better): 5 · accept x402: 0 · hosted endpoints: 0 · desk reviews by the panel: 28
- JSON: https://www.anchorterminal.com/api/v1/tools.json (list) · https://www.anchorterminal.com/api/v1/rankings.json (ranked) · https://www.anchorterminal.com/api/v1/x402.json (payable) · https://www.anchorterminal.com/api/v1/capabilities.json (by capability)
- Grades run AA, A, BB, B, C, D, E, F · methodology: https://www.anchorterminal.com/benchmark/

- Capabilities in this category: agent.framework, agent.multi-agent, agent.durable, agent.mcp-client
- https://letme.dev/agent.framework picks the top-graded tool in this list and says how to call it direct; calling through letme comes later (https://www.anchorterminal.com/letme/index.md)

## Ranking

| # | Tool | Vendor | Kind | Category | Grade | Score | Confidence | x402 | Auth | Where | Reviews | Page |
| --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- |
| 1 | OpenAI Agents SDK | OpenAI | Agent framework | Frameworks | AA | 86.5 | high | no | API key | library | 3.9/5 (8) | https://www.anchorterminal.com/tools/openai-agents-sdk.md |
| 7 | Pydantic AI | Pydantic | Agent framework | Frameworks | A | 80 | medium | no | None | library | 4/5 (8) | https://www.anchorterminal.com/tools/pydantic-ai.md |
| 45 | Agent Development Kit (ADK) | Google | Agent framework | Frameworks | BB | 74.9 | medium | no | None | library | 3/5 (8) | https://www.anchorterminal.com/tools/google-adk.md |
| 71 | Claude Agent SDK | Anthropic | Agent framework | Frameworks | BB | 72.4 | medium | no | API key | library | none | https://www.anchorterminal.com/tools/claude-agent-sdk.md |
| 95 | LangGraph | LangChain | Agent framework | Frameworks | BB | 70.6 | medium | no | None | library | 3.5/5 (2) | https://www.anchorterminal.com/tools/langgraph.md |
| 149 | CrewAI | CrewAI | Agent framework | Frameworks | B | 67 | medium | no | None | library | 3/5 (2) | https://www.anchorterminal.com/tools/crewai.md |

Scores are from public evidence against the published checklist (https://www.anchorterminal.com/benchmark/), with Performance and Task success pending. p95 latency and context cost come from our probes, which haven't run yet.

## Summaries

### 1. OpenAI Agents SDK, AA (86.5)

Multi-agent framework built on agents, hand-offs and guardrails, with sessions, tracing and human approval. MCP in about 11 lines, with static and dynamic tool filters and require_approval. Tracing on by default, with model and tool content, sent to OpenAI.

- Page: https://www.anchorterminal.com/tools/openai-agents-sdk · Markdown: https://www.anchorterminal.com/tools/openai-agents-sdk.md · JSON: https://www.anchorterminal.com/api/v1/tools/openai-agents-sdk.json
- Capabilities: agent.framework, agent.multi-agent, agent.durable, agent.mcp-client

### 7. Pydantic AI, A (80)

Typed Python agent framework for 25+ model providers, with MCP, A2A and durable execution. Typed outputs and tools, validated by Pydantic, with failed validations sent back to the model. Python only.

- Page: https://www.anchorterminal.com/tools/pydantic-ai · Markdown: https://www.anchorterminal.com/tools/pydantic-ai.md · JSON: https://www.anchorterminal.com/api/v1/tools/pydantic-ai.json
- Capabilities: agent.framework, agent.multi-agent, agent.durable, agent.mcp-client

### 45. Agent Development Kit (ADK), BB (74.9)

Google's code-first toolkit to build, evaluate and deploy agents in Python, TypeScript, Go, Java and Kotlin. Python, TypeScript, Go, Java and Kotlin. Two critical CVEs in 2026, one of them in tool confirmation itself.

- Page: https://www.anchorterminal.com/tools/google-adk · Markdown: https://www.anchorterminal.com/tools/google-adk.md · JSON: https://www.anchorterminal.com/api/v1/tools/google-adk.json
- Capabilities: agent.framework, agent.multi-agent, agent.durable, agent.mcp-client

### 71. Claude Agent SDK, BB (72.4)

Runs the Claude Code agent as a library, with its loop, built-in tools, permissions, sessions, hooks and subagents. Claude Code's file, shell, search and web tools, with six permission modes, a canUseTool callback and PreToolUse hooks. Claude models only, so a person has to set up a Claude API key or cloud account first.

- Page: https://www.anchorterminal.com/tools/claude-agent-sdk · Markdown: https://www.anchorterminal.com/tools/claude-agent-sdk.md · JSON: https://www.anchorterminal.com/api/v1/tools/claude-agent-sdk.json
- Capabilities: agent.framework, agent.multi-agent, agent.mcp-client

### 95. LangGraph, BB (70.6)

Low-level runtime for long-running, stateful agent graphs with checkpointed persistence, human-in-the-loop, memory and streaming. Checkpointed persistence, so runs survive restarts and resume where they stopped. No MCP client of its own. LangChain's langchain.mcp is in beta.

- Page: https://www.anchorterminal.com/tools/langgraph · Markdown: https://www.anchorterminal.com/tools/langgraph.md · JSON: https://www.anchorterminal.com/api/v1/tools/langgraph.json
- Capabilities: agent.framework, agent.multi-agent, agent.durable, agent.mcp-client

### 149. CrewAI, B (67)

Python framework for teams of role-playing agents in Crews, plus event-driven Flows for stateful workflows. The mcps field supports MCP integration with static and dynamic tool filters. Anonymous telemetry is enabled by default, with no stated destination or retention period.

- Page: https://www.anchorterminal.com/tools/crewai · Markdown: https://www.anchorterminal.com/tools/crewai.md · JSON: https://www.anchorterminal.com/api/v1/tools/crewai.json
- Capabilities: agent.framework, agent.multi-agent, agent.durable, agent.mcp-client

## Indexed, not reviewed (2)

Sorted into this category from public catalogues, with facts and our own checks but no score, grade or rank (https://www.anchorterminal.com/indexed/index.md).

| Listing | Kind | What it does | Why it's here |
| --- | --- | --- | --- |
| [FlurryPORT](https://www.anchorterminal.com/tools/flurryport-webhooks-and-rooms.md) | MCP server | Webhook capture and replay, delivery pipes, and signed multi-agent rooms. No signup to start. | vendor's own |
| [Gemot Deliberation Server](https://www.anchorterminal.com/tools/gemot.md) | MCP server | Deliberation primitive for multi-agent coordination: cruxes, vote clustering, consensus. | vendor's own |