KwaiKEG/KwaiAgents

Health v2 · maintenance only

Flags: no_releases no_license

Adoption not part of the score

1200 stars · 110 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

KwaiAgents is an open-source suite from Kuaishou for building generalized information-seeking agents powered by LLMs. It includes KAgentSys-Lite (an agent system), agent-tuned LLMs (KAgentLMs), a 200k+ instruction dataset (KAgentInstruct), and an evaluation benchmark (KAgentBench).

Use cases

• build an autonomous information-seeking agent with LLMs
• run an autogpt-style agent locally with open-source models
• evaluate LLM agent capabilities like planning and tool use
• fine-tune an LLM for agent behaviors with instruction data
• create a research assistant that searches and summarizes information
• benchmark agent frameworks on planning and reflection tasks

When to choose

• you want an open-source AutoGPT/BabyAGI-style agent system
• you need agent-capable open LLMs plus training and benchmark data in one project
• you are researching agent tuning methods like Meta-agent tuning

When to avoid

• you need the full-featured system with memory mechanisms described in the paper
• you want a production-ready, actively maintained agent platform
• you need a no-code or commercial agent builder

Facets

framework · maturity maintenance

agent-framework llm-inference rag search-engine machine-learning artificial-intelligence large-language-models python cross-platform autonomous-agents agent-tuning llm-agents information-seeking benchmark instruction-dataset ai-agents natural-language-processing

Member repositories

For agents

markdown · JSON · MCP: product_card(name="KwaiKEG/KwaiAgents")