# thirdlayerinc/autoagent

autonomous harness engineering

Repository: https://github.com/thirdlayerinc/autoagent
Canonical: https://ross.abutalabs.com/products/thirdlayerinc-autoagent
Language: Python
License Family: other
Last push: 2026-04-03T21:11:51+00:00

## Health v2 (maintenance only)
Score: 48/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 75, release rhythm 35, longevity 10
- inputs: {"age_days": 153, "days_push": 152, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases, young, no_license
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 4566, forks 499 (observed 2026-08-28T04:08:53.543739+00:00)

## What it is
AutoAgent is a Python-based meta-agent harness that autonomously builds and iterates on AI agent harnesses. It edits an agent's system prompt, tools, configuration, and orchestration, runs benchmark evaluations via Harbor, and hill-climbs on the benchmark score without human intervention.

## Use cases
- automatically improve an AI agent's benchmark score overnight
- let an LLM meta-agent iterate on agent prompts and tool definitions
- run autonomous agent engineering loops against Harbor-format tasks
- optimize agent orchestration and routing without manual coding
- evaluate and hill-climb agent harnesses on task suites
- bootstrap self-configuring agents from a markdown directive

## When to choose
- you want an AI to autonomously optimize an agent harness against a scored benchmark
- you already use Harbor-format evaluation tasks and Docker
- you prefer editing a single program.md directive instead of harness code
- you're experimenting with self-improving or self-configuring agent systems

## When to avoid
- you need a production-ready agent framework with a stable API and support
- you have no benchmark tasks or scoring metric to hill-climb on
- you can't provide Docker and LLM provider credentials
- you need a permissively licensed project - it currently has no license

## Facets
- artifact type: framework
- maturity: active
- function: agent-framework, llm-training, benchmarking, machine-learning, developer-tools
- domain: large-language-models, artificial-intelligence, developer-tools
- platform: python, cross-platform
- tags: meta-agent, autonomous-agents, harness-engineering, hill-climbing, self-improving-agents, agent-optimization, harbor, llm-benchmarking, ai-agents, automation, docker, linux, macos

## Member repositories
- thirdlayerinc/autoagent (main) score 48

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:08:53.543739+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 2026-08-29T18:19:58.963874+00:00, confidence not recorded.
  - readme: https://github.com/thirdlayerinc/autoagent (fetched 2026-08-28T04:08:53.543739+00:00, sha 3af589500da0)
- Data as of 2026-08-30T08:39:29.467469+00:00.