LLM 可观测性

工具排名

混合排名：实测增长优先。 原始值保留各自的数据窗口；计算变化需要 7 天内至少两次测量。

1. 1

   eval-layer

   其他

   休眠最近提交：4 个月前 · 2026-04-29

   A Claude Code skill that adds a rubric-based eval layer to any agent project. Framework-agnostic — generates rubric, test cases, judge prompt, and harness. Returns a weighted score plus a judge-leniency signal.

   HTML

   • agent-evaluation
   • anthropic
   • claude-code
   • claude-code-skill
   • evals

   github4 个月前创建实测增长打开来源 ↗

   13GitHub 星标稳定
2. 2

   evoagent-os

   智能体

   活跃最近提交：26 天前 · 2026-08-10

   Local-first control plane for durable, governed agent teams: DAG orchestration, approvals, memory, signed skills, trace contracts, and realtime.

   Python许可证: Apache-2.0

   • a2a-protocol
   • agentic-ai
   • ai-agents
   • fastapi
   • llmops

   github26 天前创建最新实测增长打开来源 ↗

   0GitHub 星标稳定
3. 3

   memroos

   智能体

   活跃最近提交：1 天前 · 2026-09-04

   MemroOS / memroos: memory OS and governance layer for AI agents, agent workflows, dispatch, proof, and context continuity.

   TypeScript

   • a2a
   • agent-governance
   • agent-memory
   • agent-orchestration
   • agent-workflows

   github5 个月前创建实测增长打开来源 ↗

   7GitHub 星标稳定
4. 4

   homestream

   其他

   活跃最近提交：53 天前 · 2026-07-14

   🔑 HomeStream · 家园·流 — 零成本自托管多Agent协作框架，通往AI世界的那把钥匙 | Zero-cost self-hosted multi-agent framework — The key to AI world

   Python许可证: MIT

   • a2a-protocol
   • agent-framework
   • ai
   • causal-memory
   • event-stream

   github57 天前创建实测增长打开来源 ↗

   0GitHub 星标稳定
5. 5

   adl-cli

   其他

   活跃最近提交：1 天前 · 2026-09-04

   A command-line tool to scaffold and manage enterprise-ready AI Agents powered by the A2A (Agent-to-Agent) protocol

   Go许可证: Apache-2.0

   • a2a
   • agents
   • ai
   • containerized
   • docker

   github13 个月前创建实测增长打开来源 ↗

   14GitHub 星标+1 (+7.7 %)
6. 6

   simple-output-styles

   技能

   活跃最近提交：今天 · 2026-09-05

   Make Claude write clearly, for everyone.

   Python许可证: 0BSD

   • claude-code
   • claude-code-plugin
   • controlled-language
   • llm-evaluation
   • output-style

   github35 天前创建实测增长打开来源 ↗

   安装 /plugin marketplace add stefanobaghino/simple-output-styles

   16GitHub 星标稳定
7. 7

   agent-mmm

   技能

   活跃最近提交：2 天前 · 2026-09-03

   Marketing Mix Model expert agent plugin for Claude Code - pymc-marketing v0.18.2+

   Python

   • claude-code-agent
   • claude-code-plugin
   • claude-code-skill
   • data-analysis
   • data-analytics

   github5 个月前创建实测增长打开来源 ↗

   安装 /plugin marketplace add Yakoub-ai/agent-mmm

   4GitHub 星标稳定
8. 8

   tulip-agents

   智能体

   活跃最近提交：9 天前 · 2026-08-26

   The agent framework where the model never holds the trigger — every consequential action clears your policy first, waits for a human when it matters, and lands on a record you can verify. Build on it, or put it around the agent you already…

   Python许可证: Apache-2.0

   • a2a
   • agent-framework
   • ai-agents
   • ai-governance
   • ai-safety

   github2 个月前创建实测增长打开来源 ↗

   1GitHub 星标稳定
9. 9

   Agents-eval

   智能体

   休眠最近提交：3 个月前 · 2026-05-17

   A Multi-Agent System (MAS) evaluation framework using PydanticAI that generates and evaluates scientific paper reviews through a three-tiered assessment approach: traditional metrics, LLM-as-a-Judge, and graph-based complexity analysis.

   Python许可证: Apache-2.0

   • a2a-protocol
   • agent-evaluation
   • agentbeats
   • ai-agents
   • benchmarks

   github19 个月前创建实测增长打开来源 ↗

   2GitHub 星标稳定
10. 10

    io.github.OtherVibes/mcp-as-a-judge

    MCP

    休眠最近提交：8 个月前 · 2025-12-15

    MCP as a Judge: a behavioral MCP that strengthens AI coding assistants via explicit LLM evaluations

    Python许可证: MIT

    mcp12 个月前创建实测增长打开来源 ↗

    安装 claude mcp add mcp-as-a-judge -- uvx mcp-as-a-judge

    17GitHub 星标稳定
11. 11

    Multi_AI_Agent_Medical_Assistant

    智能体

    休眠最近提交：9 个月前 · 2025-12-08

    An AI-powered multi-agent system that demonstrates clinical triage, OTC medication recommendations, and e-pharmacy integration for respiratory conditions. Built with modular agents that collaborate to provide safe, intelligent healthcare…

    Python

    • agent-to-agent
    • agentic-ai
    • aiagent
    • antigravity-ide
    • cursor-ai

    github9 个月前创建实测增长打开来源 ↗

    0GitHub 星标稳定
12. 12

    CustoFlow

    其他

    休眠最近提交：3 个月前 · 2026-06-04

    Multi-agent customer support system with Google ADK & Gemini 2.5 Flash Lite. Kaggle capstone demonstrating 11+ concepts. Automates 80%+ queries, <10s response time.

    Python许可证: MIT

    • a2a-protocol
    • agent-evaluation
    • google-adk
    • multi-agent-system
    • observability

    github9 个月前创建实测增长打开来源 ↗

    2GitHub 星标稳定
13. 13

    multi-agent-credit-desk

    MCP

    活跃最近提交：27 天前 · 2026-08-09

    Architecture-first Python scaffold for an auditable multi-agent corporate credit desk using A2A, MCP, deterministic credit policies, model routing, and OpenTelemetry.

    Python

    • a2a-protocol
    • agentic-ai
    • auditability
    • clean-architecture
    • credit-risk

    github51 天前创建实测增长打开来源 ↗

    0GitHub 星标稳定
14. 14

    parallel-adversarial-review

    技能

    休眠最近提交：3 个月前 · 2026-06-06

    Two Claude Code skills for adversarial code review (single-model PAR and multi-model MMAR with cross-critique to catch hallucinations), plus a fixture-based eval suite.

    Python

    • ai-agents
    • claude-code-plugin
    • code-review
    • evals
    • python

    github3 个月前创建实测增长打开来源 ↗

    安装 /plugin marketplace add prime-radiant-inc/parallel-adversarial-review

    17GitHub 星标稳定
15. 15

    mini-program-engineering-skill-suite

    技能

    活跃最近提交：4 天前 · 2026-09-01

    Mini Program Engineering Skill Suite is an Agent Skill suite for evidence‑first mini‑program development. It helps agents bring WeChat or other mini‑program projects from vague intent to reliable engineering work: project intake, product…

    Python许可证: MIT

    • agent-skills
    • ai-agents
    • alipay
    • ci-cd
    • claude-code

    github23 天前创建最新实测增长打开来源 ↗

    安装 git clone https://github.com/NocodeMrLi/mini-program-engineering-skill-suite ~/.claude/skills/mini-program-engineering-skill-suite

    45GitHub 星标+6 (+15.4 %)
16. 16

    sc-prism-releases

    技能

    活跃最近提交：2 天前 · 2026-09-03

    Run many AIs on one board and keep control of all of it. Deterministic code decides who acts — never a model. A privacy floor keeps sensitive work on your machine, your own tests decide what counts as done, and every action lands on a…

    • a2a
    • agent-orchestration
    • agent-skills
    • ai-agents
    • ai-governance

    github2 天前创建最新实测增长打开来源 ↗

    安装 git clone https://github.com/sandhusukhdeep2/sc-prism-releases ~/.claude/skills/sc-prism-releases

    1GitHub 星标稳定
17. 17

    superlog

    其他

    活跃最近提交：6 天前 · 2026-08-30

    Open-source observability tool that uses AI agents to self-heal your software

    TypeScript许可证: Apache-2.0

    • ai
    • llm
    • memory
    • observability
    • opentelemetry

    github3 个月前创建估算动量打开来源 ↗

    1 404GitHub 星标—