AI Agents¶
Personal AI agents and agentic coding assistants — self-hosted, multi-channel, and self-improving systems that operate autonomously on behalf of a user.
This domain covers the emerging category of always-on AI agents that go beyond chat interfaces to execute tasks, manage workflows, and learn from experience. The space exploded in late 2025 when open-source projects demonstrated that a single agent could unify messaging platforms, execute shell commands, automate browsers, and improve its own skills over time.
Landscape (2026)¶
The AI agent space has settled into three philosophical camps:
| Philosophy | Representative | Focus |
|---|---|---|
| Deep coding | Claude Code, Codex, Cursor | Terminal-native code generation, refactoring, and CI/CD |
| Universal automation | OpenClaw, nanobot | Cross-platform personal assistant across 24+ messaging channels |
| Self-improving | Hermes Agent | Persistent memory, autonomous skill creation, evolutionary optimization |
Not Mutually Exclusive
Many users run Claude Code + OpenClaw (coding + personal automation) or Claude Code + Hermes (development + research). The agents complement rather than compete.
Key Projects¶
Core Agents¶
- OpenClaw — Open-source personal AI assistant (360k+ GitHub stars). Multi-channel gateway connecting 24+ messaging platforms to any LLM. Created by Peter Steinberger, MIT licensed.
- Hermes Agent — Self-improving AI agent by Nous Research (48k+ GitHub stars). Built-in learning loop that creates and evolves skills from experience. MIT licensed.
Claw Family Ecosystem¶
The "Claw" family is the ecosystem of projects built around or extending OpenClaw:
| Project | Maintainer | Purpose |
|---|---|---|
| NemoClaw | NVIDIA | Secure OpenClaw runtime inside NVIDIA OpenShell with managed inference |
| claworc | Community | Multi-instance orchestration dashboard for OpenClaw |
| AutoResearchClaw | AIMING Lab | 23-stage autonomous research pipeline from idea to paper |
| openclaw-termux | Community | OpenClaw on Android via Flutter + built-in terminal |
| Paperclip | PaperclipAI | Enterprise management layer on top of OpenClaw |
| ClawTeam | Community | Multi-agent swarm coordination with OpenClaw as default agent |
| clawdbot | Legacy | Original name for OpenClaw before the Anthropic trademark dispute |
Supporting Infrastructure¶
| Project | Maintainer | Purpose |
|---|---|---|
| OpenViking | ByteDance | Context database for AI agents (targets OpenClaw) |
| OpenSpace | HKU DS Lab | One-command evolution for OpenClaw, Claude Code, Codex, Cursor |
| Lobster | OpenClaw | Typed workflow shell for deterministic multi-step pipelines |
Comparisons¶
- OpenClaw vs Hermes Agent vs Claude Code — three agents, three philosophies
Sources¶
- OpenClaw GitHub — main repository (360k+ stars)
- OpenClaw Wikipedia
- OpenClaw Documentation
- Hermes Agent GitHub — main repository (48k+ stars)
- Hermes Agent Documentation
- NemoClaw Developer Guide
- NVIDIA NemoClaw Product Page
- AutoResearchClaw GitHub
- The New Stack — Persistent AI Agents Compared
- OpenClaw Architecture Deep Dive (Substack)
- OpenClaw Architecture (DeepWiki)
Questions¶
Open¶
- How does OpenClaw's security posture evolve after the 9 CVEs disclosed in March 2026 — is the ClawHub skill marketplace trustworthy?
- Will Hermes Agent's self-evolution approach (DSPy + GEPA) become the standard pattern for agent improvement, or is it a research curiosity?
- Does NemoClaw's OpenShell sandboxing model scale for enterprise deployments, or is it still early-preview quality?
- What is the long-term governance model for OpenClaw after Steinberger joined OpenAI and handed off to the non-profit foundation?
- Can Lobster's workflow engine compete with LangGraph and similar DAG-based orchestration frameworks?
Answered¶
- Q: What is the "Claw" family? — Ecosystem of projects extending OpenClaw: NemoClaw (NVIDIA security), claworc (orchestration), AutoResearchClaw (research), openclaw-termux (Android), Paperclip (enterprise), ClawTeam (swarm coordination).
- Q: Is OpenClaw related to Claude/Anthropic? — Indirectly. Originally named "Clawd" after Anthropic's Claude, it was renamed after trademark complaints. No technical or organizational relationship.
- Q: Which agent for coding vs personal automation? — Claude Code for deep coding, OpenClaw for universal messaging/automation, Hermes Agent for self-improving research workflows. Many users combine two.