YeQing17-2026/OmniAgent

Health v2 · maintenance only

Flags: no_releases young no_license

Adoption not part of the score

2549 stars · 383 forks observed · 2026-08-28

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

OmniAgent is an open-source Python agent framework that self-evolves across skills, context, memory, and its underlying model during interaction, inspired by OpenClaw. It pairs this evolution with dynamic security hardening via a Hyper-Harness execution scaffold and a dual-layer Deep Reflexion architecture.

Use cases

• build a self-evolving ai agent that learns from interactions
• run a coding assistant that remembers my project and avoids past mistakes
• safely execute shell commands with sandboxing and human approval
• deploy a chatbot to cli, web, discord, and telegram from one codebase
• research assistant that searches the web and remembers what worked
• agent framework with online reinforcement learning for the brain model

When to choose

• you want an agent whose skills, memory, and context improve in real time during use
• you need layered safety controls (llm review, policy engine, approval, sandbox) for agent command execution
• you want one agent deployable across cli, web ui, feishu, discord, and telegram
• you're comparing against OpenClaw or Hermes and want faster skill evolution and lower token cost

When to avoid

• you need a simple, static agent with predictable, unchanging behavior
• you require a permissively licensed dependency (GPL-3.0)
• you need a mature, long-proven framework - the project is new and docs are still in progress
• you don't want a self-modifying agent touching your model or context

Facets

framework · maturity active

agent-framework llm-inference security chatbot machine-learning cli webhook artificial-intelligence large-language-models security developer-tools python cli cross-platform self-hosted self-evolving-agent omni-evolve proactive-memory skill-evolution context-evolution online-reinforcement-learning hyper-harness deep-reflexion sandboxing multi-channel-deployment openclaw-inspired ai-agents automation discord telegram

Member repositories

For agents

markdown · JSON · MCP: product_card(name="YeQing17-2026/OmniAgent")