realvco Docs

Hermes-Agent Overview (pre-installed on Ada / Vi)

realvco pre-installs Hermes-Agent (open-source AI agent software) in the Ada (HM-2) and Vi (HM-3) containers.

For deep Hermes-Agent learning, see the upstream manual:

  • 📘 Hermes-Agent official docs: see the upstream resource links in Hermes-Agent release notes per version
  • 💻 Hermes-Agent open-source repo: refer to the upstream tracked by Vi’s current container version on your mVPS

This page only briefly introduces Hermes-Agent’s role on the realvco platform + our specific configuration. All general Hermes-Agent knowledge (commands, APIs, advanced usage) defers to the upstream documentation.

Hermes-Agent in one sentence

Hermes-Agent = lightweight, fast open-source AI agent software. Designed for low latency, low cost, simple Q&A, complementing OpenClaw.

Features:

  • Faster response than OpenClaw (instant on simple questions)
  • 30-50% lower token cost than OpenClaw
  • Strong English performance
  • Better for FAQ-level questions, not for deep reasoning

Upgrade highlights (latest: v0.14.0, 2026-05-16)

Ada / Vi’s pre-installed Hermes-Agent keeps updating. These are the highlights you can see / use in the admin-panel (examples highlight Vi; Ada gets the same capabilities):

  • 22 messaging platforms — since v0.12, progressively added Microsoft Teams, Tencent Yuanbao, Google Chat, LINE, SimpleX Chat (LINE is especially popular in Taiwan / Japan / Korea); all connectable in platform connection settings
  • Autonomous Curator — Vi maintains its own skill library (auto grade / consolidate / prune every 7 days)
  • Native Spotify integration (7 tools: play / search / queue / playlists / device control)
  • Google Meet integration (join meetings / live transcription / speak on your behalf / post-meeting follow-up)
  • Major TUI upgrade: LaTeX math rendering, /resume to switch sessions, /reload env hot-reload, auto-resume, mouse-wheel scrolling — full guide at Hermes TUI via web terminal

v0.13 / v0.14 also bring a batch of engine-internal improvements (lighter footprint, faster cold start, more model providers, etc.). These are general Hermes-Agent capabilities — realvco has already upgraded you automatically; details defer to the upstream documentation (see callout at top).

Full changelog: Hermes v2026.4.30 release notes (v0.13 / v0.14 notes being compiled).

Comparison: What Rose, Ada and Vi Are Each Good At.

Hermes-Agent’s role on the realvco platform

Every realvco mVPS pre-configures 2 Hermes-Agent containers:

ContainerDefault roleDefault permissionStrengths
HM-2 (Ada)High-volume support, creative writing, knowledge Q&AContainer isolatedHigh-volume daily conversation, strong English, cost efficiency
HM-3 (Vi)Platform breadth, English scenarios, self-maintaining skill libraryContainer isolated22 messaging platforms, Curator auto-curation, fast replies

Runs in parallel and independent of the OpenClaw container (OC-1). Ada and Vi both come with Hermes-Agent and have similar capabilities — a common split is “Ada handles daily support and high-volume conversation; Vi focuses on multi-platform reach and skill-library curation” — or run the same kind of work in parallel on both (they don’t interfere).

realvco-specific configurations

The following are added by realvco on top of Hermes-Agent:

1. Pre-installed

Hermes-Agent open-source version requires you to compile + configure. realvco gives you a running container directly.

2. Dashboard UI

Ada / Vi share the same Dashboard as OC-1 — in HM-2 / HM-3 tabs you can:

  • View Ada / Vi live status
  • Pause / resume
  • Edit persona / switch model
  • Track token usage
  • Upgrade software version

3. Platform connection integration

Ada / Vi can both connect to Telegram / WhatsApp / LINE / Discord / Slack — same convenience as Rose.

4. Container isolation from OpenClaw

Ada / Vi issues don’t affect Rose. The 3 agents back each other up.

5. Independent upgrade pool

Ada / Vi use an independent Hermes version pool (decoupled from OpenClaw), can upgrade independently. See Version Upgrade.

6. Hermes’s database is auto-guarded

Hermes-Agent (Ada / Vi) writes its own database (conversation sessions, skill library; from v0.12 the built-in Curator tidies the skill library every 7 days). realvco guards that database with three layers:

  • Curator can’t touch the core — when Curator auto-tidies, bundled / built-in skills cannot be modified by it; it only reorganizes your custom skills, and every run produces a report you can review
  • Same backup as OpenClaw — Ada / Vi’s database is auto-backed-up daily at 05
    , “Backup Now” anytime, 7-day rolling. The protection is exactly as strong as Rose’s — not lesser
  • Worst case, restore point backstop — if the skill library really gets scrambled, restore the whole container from a 7-day backup, or roll the whole host back to before

Hermes-Agent use cases (conceptual)

Ada / Vi are suited for:

  • High-volume repetitive simple questions (FAQ-level)
  • English-user-dominated scenarios
  • Many messaging platforms (supports 22, incl. Microsoft Teams, Tencent Yuanbao, Google Chat, LINE, SimpleX Chat)
  • Wanting a self-maintaining skill library (Curator tidies automatically, one less thing to manage)
  • Cost-sensitive high-volume scenarios

Specific case studies and examples: see Hermes-Agent upstream documentation (see callout at top).

vs OpenClaw

See What Rose, Ada and Vi Are Each Good At.

In a line:

  • Hermes-Agent (Ada / Vi) = fast, low cost, platform breadth, self-maintaining skill library
  • OpenClaw (Rose) = deep reasoning, cross-tool chaining, brand voice

English SaaS, high-volume simple customer questions, and workloads needing many messaging platforms are a great fit for Ada / Vi. (Rose comes with OpenClaw pre-installed, Ada / Vi with Hermes-Agent, managed uniformly by realvco.)

What to learn deeper

realvco operation layer (in this docs site):

Hermes-Agent general knowledge (go to upstream docs):

  • API / tool syntax → upstream docs
  • Advanced configuration → upstream GitHub repo
  • Model selection → upstream + realvco model selection comparison