Migrate from OpenClaw

Original Article ↗ Hacker News Discussion ↗

Migrating from OpenClaw to Hermes Agent prompts wider questions about whether these heavyweight “agentic assistant” stacks meaningfully improve on simpler setups like Claude Code, Codex, or custom scripts. Commenters contrast OpenClaw’s power and complexity with Hermes’ more polished, opinionated approach, but many report instability, bloat, and unclear real-world payoff beyond automated marketing or personal dashboards. Concerns over trust—ranging from plagiarism allegations and astroturfed hype to default network routing and preinstalled skills—lead some to favor minimal, self-built agents or lighter frameworks that expose LLM capabilities without the overhead of full-blown agent platforms.

Concerns and Controversies Around Hermes / NousResearch

  • Several comments reference a plagiarism/similarity accusation between Hermes Agent and another open-source agent system.
  • One side sees it as a serious “debacle”: original GitHub issue text was replaced with a single character, comments were deleted, and users blocked, which is viewed as evasive and unprofessional.
  • Others argue the claims are weak: core ideas (memory, skills, evolution of agents) are generic and widely implemented, with no concrete code-level evidence presented.
  • There is mention of Hermes default installs previously routing web search through a specific provider; critics disliked the communication and perceived data-leak risk. Supporters say it was just a free default search provider that was removed once criticized.
  • Broader discussion touches on AI, copyright, and how trivial copying may erode traditional IP norms; some see this as ethically troubling but practically inevitable.

Hermes vs OpenClaw (and Similar Agents)

  • Hermes is described as more polished, focused, and friendlier for less-technical users.
  • OpenClaw is seen as more powerful and deeper, but also messy, unreliable, and hard to configure. Some report broken onboarding, confusing UI, and frequent failures; others say it works well for them and has strong CLI onboarding, especially for non-technical users.
  • Some feel Hermes ships with too many pre-enabled “skills,” wasting context and adding clutter; recent cleanup efforts are acknowledged but not universally convincing.

Use Cases and Reported Successes

  • Reported real-world uses include:
    • Lead generation and outreach pipelines.
    • Company data access for nontechnical staff via Slack, with connections to analytics and billing systems.
    • Personal automations: email triage, spam filtering, web monitoring, notifications, ticket/PR workflows, and follow-up nudging.

Skepticism About “Agentic” Systems

  • Many participants struggle to find compelling use cases beyond what cron jobs, scripts, and dedicated coding agents already do.
  • Some view OpenClaw and similar tools as toys, “larping,” or hype vehicles for technophiles and influencers, with unclear economic value for end users.
  • A recurring theme: anyone with a serious, well-understood use case often ends up building a custom, slimmer harness (e.g., simple LLM + memory + tooling) instead of relying on large, general-purpose agent frameworks.