# facebookresearch/HyperAgents

Self-referential self-improving agents that can optimize for any computable task

Repository: https://github.com/facebookresearch/HyperAgents
Canonical: https://ross.abutalabs.com/products/hyperagents
Language: Python
License: NOASSERTION
License Family: other
Last push: 2026-07-31T16:56:22+00:00

## Health v2 (maintenance only)
Score: 57/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 95, release rhythm 35, longevity 11
- inputs: {"age_days": 167, "days_push": 33, "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 2697, forks 359 (observed 2026-08-28T04:07:11.234745+00:00)

## What it is
A research framework from Meta AI implementing self-referential, self-improving LLM agents that can optimize their own code for arbitrary computable tasks. It provides a meta agent and task agent loop that iteratively generates and refines agent implementations across multiple domains.

## Use cases
- run self-improving AI agents that optimize arbitrary tasks
- research self-referential agent architectures
- experiment with LLM agents that rewrite their own code
- reproduce HyperAgents paper experiments
- benchmark meta-agent performance across domains

## When to choose
- you're researching self-improving or self-referential agent systems
- you want to reproduce or extend the HyperAgents paper
- you need a framework for agents that optimize their own implementations

## When to avoid
- you need production-ready, safety-audited agent deployment
- you require a permissive commercial license (CC BY-NC-SA 4.0)
- you can't safely sandbox execution of untrusted model-generated code

## Facets
- artifact type: library
- maturity: active
- function: agent-framework, llm-inference, machine-learning
- domain: artificial-intelligence, large-language-models
- platform: python, cli
- tags: self-improving-agents, meta-agent, research-code, llm-agents, non-commercial-license, code-generation, ai-agents, research, linux, docker

## Member repositories
- facebookresearch/HyperAgents (main) score 57

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:07:11.234745+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 2026-08-30T02:16:12.907572+00:00, confidence not recorded.
  - readme: https://github.com/facebookresearch/HyperAgents (fetched 2026-08-28T04:07:11.234745+00:00, sha 897fa1452123)
- Data as of 2026-08-30T08:39:29.467469+00:00.