# mozilla-ai/any-agent

A single interface to use and evaluate different agent frameworks

Repository: https://github.com/mozilla-ai/any-agent
Canonical: https://ross.abutalabs.com/products/any-agent
Homepage: https://docs.mozilla.ai/any-agent
Language: Python
License: Apache-2.0
License Family: permissive
Topics: agent-evaluation, agents, ai, a2a, mcp
Last push: 2026-08-01T00:52:22+00:00

## Health v2 (maintenance only)
Score: 71/100 (v2, computed 2026-09-03T02:39:23.370411+00:00)
- activity 95, release rhythm 59, longevity 37
- inputs: {"age_days": 526, "days_push": 33, "days_rel": 196, "gap_med": 2.0, "n_releases_24m": 73}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1197, forks 95 (observed 2026-08-28T04:03:57.527690+00:00)

## What it is
any-agent is a Python library providing a single unified interface for building, serving, and evaluating AI agents across multiple agent frameworks such as Agno, Google ADK, LangChain, LlamaIndex, OpenAI Agents SDK, and smolagents. It offers framework-agnostic APIs, standardized OpenTelemetry tracing, built-in LLM-as-a-judge evaluation, and serving via A2A or MCP protocols.

## Use cases
- switch between agent frameworks with a single parameter change
- evaluate agent performance with llm-as-a-judge
- get unified opentelemetry traces across agent frameworks
- serve an agent via a2a or mcp protocol
- use an agent as a tool for other agents
- compare agent frameworks under one api
- build a first ai agent without locking into one framework

## When to choose
- you need to run or evaluate agents across multiple frameworks under one API
- you want standardized tracing and evaluation tooling for agents
- you need to serve agents via A2A or MCP and compose them as tools

## When to avoid
- you only need a minimal core agent loop - use mozilla-ai-tinyagent instead
- you are starting a new project - the project is in soft deprecation with no new features planned
- you are committed to a single agent framework and don't need cross-framework abstraction

## Facets
- artifact type: library
- maturity: maintenance
- function: agent-framework, llm-inference, mcp, tracing, testing
- domain: large-language-models, artificial-intelligence, developer-tools
- platform: python, cross-platform
- tags: agent-evaluation, a2a, multi-framework, llm-as-a-judge, opentelemetry, soft-deprecated, ai-agents

## Member repositories
- mozilla-ai/any-agent (main) score 71

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:03:57.527690+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T06:21:28.291049+00:00, confidence not recorded.
  - readme: https://github.com/mozilla-ai/any-agent (fetched 2026-08-28T04:03:57.527690+00:00, sha 23648d7e0ef8)
  - homepage: https://docs.mozilla.ai/any-agent (fetched 2026-08-29T12:29:00.464097+00:00, sha 26225bd8c32c)
  - registry_pypi: https://pypi.org/pypi/any-agent/json (fetched 2026-08-29T12:29:00.491341+00:00, sha 4c047df62dab)
- Data as of 2026-08-30T08:39:29.467469+00:00.