The Index

Every entry verified live, scored against the public rubric, and timestamped. Sorted by overall score, never by stars.

metrics fetched 2026-07-07 · methodology v1.0

24 of 46 entries

promptfoo/promptfooA-Tier

Test harness for prompts, agents, and RAG pipelines: declarative test cases, assertions, matrix comparisons, and red-team scanning, runnable…

83/100 MIT active (commit 0d before fetch)
generic
langchain-ai/langgraphA-Tier

The most adopted graph-based agent orchestration framework: workflows as explicit state machines with checkpointing, human-in-the-loop…

82/100 MIT active (commit 1d before fetch)
generic
pydantic/pydantic-aiA-Tier

An agent framework built on type validation: structured outputs, dependency injection, and tool schemas checked the way Pydantic checks…

82/100 MIT active (commit 0d before fetch)
generic
anthropics/skillsA-Tier

Anthropic's first-party public skill collection: the de facto reference for SKILL.md structure, frontmatter conventions, and…

81/100 no license active (commit 6d before fetch)
claude-codegeneric
confident-ai/deepevalA-Tier

A pytest-style LLM evaluation framework: metric classes for correctness, hallucination, RAG quality, and agent task completion, written as…

80/100 Apache-2.0 active (commit 1d before fetch)
generic
langfuse/langfuseA-Tier

Open-source LLM engineering platform: tracing, evals, prompt management, and metrics for agent systems. The observability layer most stacks…

80/100 NOASSERTION active (commit 0d before fetch)
generic
microsoft/playwright-mcpA-Tier

Browser automation as agent tools, built on Playwright's accessibility tree rather than screenshots: structured, deterministic web…

80/100 Apache-2.0 active (commit 8d before fetch)
claude-codecursorcodexcopilotgeneric
modelcontextprotocol/serversA-Tier

The reference implementations of the Model Context Protocol: the official server collection that defines what a well-built agent tool looks…

79/100 NOASSERTION active (commit 1d before fetch)
claude-codecursorcodexcopilotgeneric
upstash/context7A-Tier

An MCP server that feeds agents current, version-specific library documentation at generation time, attacking the stale-training-data class…

79/100 MIT active (commit 0d before fetch)
claude-codecursorcodexcopilotgeneric
openai/openai-agents-pythonA-Tier

OpenAI's first-party multi-agent framework: agents, handoffs, guardrails, and tracing in a deliberately small API surface.

79/100 MIT active (commit 0d before fetch)
generic
github/github-mcp-serverA-Tier

GitHub's official MCP server: repos, issues, PRs, actions, and code scanning exposed as agent tools with GitHub's own auth and permission…

79/100 MIT active (commit 0d before fetch)
claude-codecursorcodexcopilotgeneric
mem0ai/mem0A-Tier

A universal memory layer for AI agents: extraction, storage, and retrieval of user and session memories with SDKs across the major agent…

78/100 Apache-2.0 active (commit 0d before fetch)
generic
google/adk-pythonA-Tier

Google's Agent Development Kit: code-first agent construction with built-in evaluation tooling and deployment paths into the Google Cloud…

78/100 Apache-2.0 active (commit 0d before fetch)
generic
K-Dense-AI/scientific-agent-skillsA-Tier

140 domain skills plus 100+ database connectors that turn a coding agent into a working research assistant for biology, chemistry, medicine,…

77/100 MIT active (commit 1d before fetch)
claude-codecursorcodexgeneric
letta-ai/lettaA-Tier

A platform for stateful agents with self-editing memory, descended from the MemGPT research line: agents manage their own memory blocks…

77/100 Apache-2.0 active (commit 4d before fetch)
generic
oraios/serenaA-Tier

An MCP toolkit that gives coding agents semantic code retrieval and editing: language-server-grade symbol operations instead of…

77/100 MIT active (commit 1d before fetch)
claude-codecursorcodexgeneric
guardrails-ai/guardrailsA-Tier

Input/output guarding for LLM applications: validator library plus enforcement framework, with correction strategies when checks fail.

77/100 Apache-2.0 active (commit 7d before fetch)
generic
crewAIInc/crewAIA-Tier

Role-based multi-agent orchestration: crews of agents with roles, goals, and tasks, optimized for approachability over control-flow…

76/100 MIT active (commit 0d before fetch)
generic
microsoft/autogenA-Tier

Microsoft's agentic programming framework: conversation-based multi-agent patterns from the original research lineage, in transition toward…

75/100 CC-BY-4.0 commit 83d before fetch
generic
getzep/zepA-Tier

A temporal knowledge-graph memory layer for agents: facts extracted from conversations with validity intervals, so recall can answer what…

74/100 Apache-2.0 active (commit 1d before fetch)
generic
topoteretes/cogneeA-Tier

An open-source memory platform that builds knowledge graphs plus embeddings from agent data, aiming at reasoning-grade recall rather than…

74/100 Apache-2.0 active (commit 0d before fetch)
generic
wshobson/agentsA-Tier

A multi-harness plugin marketplace of role-specialized subagents (reviewer, architect, security, docs) installable across Claude Code, Codex…

72/100 MIT active (commit 1d before fetch)
claude-codecodexcursorcopilotgeneric
basicmachines-co/basic-memoryA-Tier

Local-first agent memory over plain Markdown files with an MCP interface: knowledge persists as human-readable notes the agent reads and…

71/100 AGPL-3.0 active (commit 0d before fetch)
claude-codecursorgeneric
eyaltoledano/claude-task-masterA-Tier

An AI task-management loop that decomposes a spec into ordered, dependency-aware tasks and drives an agent through them one at a time across…

71/100 no license commit 70d before fetch
cursorwindsurfclaude-codegeneric