# TencentCloudADP/youtu-agent

A simple yet powerful agent framework that delivers with open-source models

Repository: https://github.com/TencentCloudADP/youtu-agent
Canonical: https://ross.abutalabs.com/products/youtu-agent
Homepage: https://tencentcloudadp.github.io/youtu-agent/
Language: Python
License: NOASSERTION
License Family: other
Topics: agent-framework, agents, openai-agents, python
Last push: 2026-03-21T07:18:28+00:00

## Health v2 (maintenance only)
Score: 57/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 73, release rhythm 54, longevity 26
- inputs: {"age_days": 377, "days_push": 165, "days_rel": 307, "gap_med": 11, "n_releases_24m": 6}
- flags: no_license
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 4604, forks 477 (observed 2026-08-28T04:08:54.907853+00:00)

## What it is
Youtu-Agent is a Python framework for building, running, and evaluating autonomous LLM agents, supporting ReAct-style single agents and plan-and-execute multi-agent systems. It works with open-weight models like DeepSeek-V3, and includes automated agent generation, experience-based learning, and end-to-end RL training pipelines.

## Use cases
- build autonomous ai agents with open-source llms
- run deep research agents on gaia benchmark
- train agents with reinforcement learning
- automatically generate agent tools and prompts
- evaluate llm agents on web qa benchmarks
- create multi-agent plan-and-execute workflows
- improve agent performance without fine-tuning

## When to choose
- you want state-of-the-art open-source agent baselines on benchmarks like GAIA or WebWalkerQA
- you need end-to-end RL training for agents at scale
- you prefer open-weight models like DeepSeek over proprietary APIs
- you want automated generation of agent configs, prompts, and tool code

## When to avoid
- you need a simple chatbot with minimal setup
- you depend exclusively on closed-source model ecosystems like OpenAI Assistants
- you need a production-ready no-code agent platform for non-developers

## Facets
- artifact type: framework
- maturity: active
- function: agent-framework, llm-inference, llm-training, web-scraping, data-science, benchmarking
- domain: large-language-models, artificial-intelligence, reinforcement-learning, developer-tools
- platform: python
- tags: deep-research, gaia-benchmark, rl-training, multi-agent, open-source-models, tool-synthesis, hydra-config, ai-agents, linux, macos, docker, gpu

## Member repositories
- TencentCloudADP/youtu-agent (main) score 57

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:08:54.907853+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 2026-08-29T18:19:45.095235+00:00, confidence not recorded.
  - readme: https://github.com/TencentCloudADP/youtu-agent (fetched 2026-08-28T04:08:54.907853+00:00, sha c095e5c826fa)
  - homepage: https://tencentcloudadp.github.io/youtu-agent/ (fetched 2026-08-29T09:04:54.138978+00:00, sha d53478ed024a)
  - registry_pypi: https://pypi.org/pypi/youtu-agent/json (fetched 2026-08-29T09:04:54.148095+00:00, sha f8ab3cc83567)
- Data as of 2026-08-30T08:39:29.467469+00:00.