# google/ax

An open source distributed agent runtime

Repository: https://github.com/google/ax
Canonical: https://ross.abutalabs.com/products/google-ax
Homepage: https://agentexecutor.io
Language: Go
License: Apache-2.0
License Family: permissive
Last push: 2026-08-20T05:56:51+00:00

## Health v2 (maintenance only)
Score: 80/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 98, release rhythm 97, longevity 11
- inputs: {"age_days": 156, "days_push": 13, "days_rel": 20, "gap_med": 11.0, "n_releases_24m": 5}
- flags: young
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1970, forks 116 (observed 2026-08-28T04:06:00.798066+00:00)

## What it is
AX (Agent Executor) is a distributed agent runtime from Google that provisions isolated, suspendable/resumable environments to execute AI agents and harnesses. It provides durable event logging, automatic failure recovery, and execution resumption, with first-class Kubernetes support and native MCP/A2A protocol interoperability.

## Use cases
- run long-running autonomous AI agents reliably in distributed deployments
- resume agent executions after failures or interruptions
- deploy agents as isolated actors on Kubernetes
- audit and observe agent trajectories and execution state
- orchestrate multi-agent systems with MCP and A2A protocols
- bring custom LLM harnesses, tools, and skills to a managed runtime

## When to choose
- you need durable, resumable execution for long-running autonomous agents
- you deploy agents across distributed or Kubernetes-based infrastructure
- you want isolated, multi-tenant agent execution with full audit trails
- you need MCP/A2A interoperability with a model- and harness-agnostic runtime

## When to avoid
- you need a stable, production-hardened system - AX is in early development with expected breaking changes
- you want a simple single-process agent framework without distributed runtime overhead
- you need broad community contributions right now - external PRs are temporarily paused
- your use case is basic chatbot or prompt orchestration rather than autonomous execution

## Facets
- artifact type: framework
- maturity: experimental
- function: agent-framework, mcp, workflow-automation, container-orchestration, monitoring
- domain: large-language-models, microservices, infrastructure-as-code
- platform: go, cloud, self-hosted, cli, windows
- tags: agent-runtime, distributed-agents, resumable-execution, a2a, event-log, single-writer, harness, google, ai-agents, devops, kubernetes, docker, linux, macos

## Member repositories
- google/ax (main) score 80

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:06:00.798066+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 2026-08-30T03:05:08.025787+00:00, confidence not recorded.
  - readme: https://github.com/google/ax (fetched 2026-08-28T04:06:00.798066+00:00, sha e1a38e6d3955)
  - homepage: https://agentexecutor.io (fetched 2026-08-29T10:44:37.536823+00:00, sha e35d6a8c4035)
- Data as of 2026-08-30T08:39:29.467469+00:00.