8-min read · Curated by an AI Systems Architect
Focus: Hermes v0.17.0 · Background Agents · Multi-Platform Reach
Hermes Agent just dropped v0.17.0 — codenamed "The Reach Release." The numbers are staggering: 1,475 commits, 800 merged PRs, 230,000 new lines of code. The official line sums it up perfectly: "v0.16.0 put Hermes on your desktop. v0.17.0 is about how far that reach extends."
I read through the entire changelog so you don't have to. Here's the thing — most of these changes aren't about adding more AI capabilities. They're about making Hermes show up where you need it and work while you do other things. That's the theme.
If you're an everyday user who uses Hermes for coding, research, or just automating life, here are the 7 features in v0.17.0 you'll actually use — with step-by-step instructions.
1. Background Sub-Agents — Stop Waiting for AI to Finish
What it is: You can now delegate tasks to sub-agents that run in the background. You give Hermes a task, it says "got it," and you go back to whatever you were doing. Results arrive in your chat when they're ready.
Why it matters: Before v0.17.0, if you asked Hermes to "research the latest 10 AI papers and summarize each," you'd stare at the screen for 5 minutes. Now it runs in the background while you keep working.
How to use it:
In any conversation, just tell Hermes:
"In the background, search for the top 5 AI agent papers this week and write me a summary."
Hermes automatically calls delegate_task(background=true). You'll see an immediate confirmation, then keep chatting. When the task finishes, the results appear in the same conversation.
In messaging apps (Telegram, Discord, WhatsApp):
/background Check all my servers and flag any that are down
Hermes replies immediately:
🔄 Background task started: "Check all my servers..."
Task ID: bg_143022_a1b2c3
Pro tip: You can set a cheaper model for background tasks in ~/.hermes/config.yaml:
delegation:
model: "google/gemini-flash-2.0"
provider: "openrouter"
💡 Best use cases:
- Morning research briefings while you make coffee
- Running test suites in the background while coding
- Multi-tasking: spawn 3 background agents for papers, news, and data — results arrive separately
— Hermes Docs · The Reach Release Changelog
🔗 Hermes Documentation — Background Sessions · Hermes v0.17.0 Release
2. Automation Blueprints — Cron Without Learning Cron
What it is: A form-based automation builder in the dashboard. Pick a task template, answer a few questions (what time? where to send?), and save. No cron syntax required.
Why it matters: I've been using cron for years and I still need to Google the syntax sometimes. Automation Blueprints remove that friction entirely. Anyone can set up daily AI briefings, weekly reports, or health checks in 30 seconds.
How to use it:
- Open the Hermes Dashboard (
hermes dashboard) - Find the Automation section
- Pick a blueprint from the dropdown — e.g., "Daily AI News Briefing"
- Answer the prompts: time → 7:30 AM, delivery channel → Telegram
- Save
That's it. Next morning at 7:30, Telegram gets your briefing.
Example blueprints available:
- Daily news briefing
- Server health check
- Weekly report generation
- Scheduled code review
You can also set one-time reminders conversationally — just say "Remind me in 1 hour to join the product review meeting" and Hermes creates it for you.
— Hermes Docs
3. Desktop App — It Feels Like Real Software Now
What it is: v0.16.0 introduced the desktop app, but it was more of a preview. v0.17.0 adds the polish that turns "functional" into "pleasant."
Why it matters: The little things add up. Here's what you get:
| Feature | How to Use It |
|---|---|
| Rebindable keyboard shortcuts | Settings → Keyboard → remap common actions |
| Native system notifications | Enable per-category in Settings → Notifications |
| Sub-agent monitor window | When background tasks run, a live window shows tool calls in real time |
| VS Code theme import | Settings → Themes → Import from VS Code. Your editor colors sync to Hermes |
| Per-thread drafts | Switch conversations freely — drafts never get lost |
| Resizable terminal panel | Drag the terminal border like VS Code |
The VS Code theme import is the killer feature for me. I spent months getting my editor colors exactly right — now I don't have to learn a new color scheme. That kind of comfort matters more than any single functional feature.
New languages in the desktop UI: Japanese and Traditional Chinese are now supported.
— Hermes v0.17.0 Changelog
4. Image-to-Image Editing — AI Photo Editing in Plain English
What it is: The image_generate tool now supports editing existing images — not just generating new ones from scratch. Change colors, remove backgrounds, turn sketches into renders.
Why it matters: You don't need to open Photoshop or Canva for quick edits. Drag an image into Hermes and describe what you want in plain English.
How to use it:
Drag an image into your conversation and say:
"Remove the background from this image, keep only the subject."
Or:
"Turn this sketch into a photorealistic render."
Or:
"Change this logo's color to dark blue."
Hermes passes your image + instruction to the image generation backend and returns the edited result. Works through Nous Portal with no additional setup.
💡 Most practical use case: Article cover images. Draft a cover, ask Hermes to "make the title text larger," "darken the background" — two rounds and you're done.
— Hermes v0.17.0 Changelog
5. iMessage & WhatsApp — No Mac Relay Needed
What it is: Two new messaging platform integrations that remove friction:
- iMessage via Photon Spectrum — No Mac relay required anymore. Any device can connect.
- WhatsApp Business Cloud API — Official Meta API, no bridge process needed.
- Telegram Rich Text — Messages now render with Bot API 10.1 formatting (code blocks, bold, lists).
Why it matters: Hermes follows you to the apps you already use. You don't adapt to Hermes — Hermes adapts to you.
How to set it up:
One command for everything:
hermes gateway setup
An interactive menu appears. Use arrow keys to select:
-
iMessage → choose Photon Spectrum, run
hermes photon login, authenticate with device code - WhatsApp → choose WhatsApp Cloud API, enter your Business API credentials
- Telegram → enter your bot token
After setup:
hermes gateway start
Now you can message Hermes from any of these apps, and it remembers your identity across platforms.
⚠️ Security reminder: Set ALLOWED_USERS environment variables or use DM pairing so only authorized users can talk to Hermes.
— Hermes Docs
🔗 Hermes Messaging Gateway · iMessage Setup · WhatsApp Cloud API Setup
6. Curator Cost Savings — No More Silent Token Burns
What it is: Hermes's Curator (skill manager) previously called LLMs during routine cleanup to decide "which skills to merge." That burned tokens — and cost money — on every cleanup cycle.
v0.17.0 disables consolidation by default. The Curator now only performs pure cleanup (no LLM calls, zero token cost). Skill consolidation is still available — just opt in.
Why it matters: If you use paid APIs like OpenAI, Curator was silently spending money you didn't know about. Now it doesn't. Zero action required — just update to v0.17.0.
If you want consolidation back:
hermes curator run --consolidate
Or permanently enable in ~/.hermes/config.yaml:
curator:
consolidate: true
— Hermes v0.17.0 Changelog
7. Other Notable Improvements
Quick hits worth knowing about:
-
Dashboard Profile Editor — Create and edit profiles in the browser. No more hand-editing
config.yaml. - Skills Hub Redesign — Category recommendations, install previews, and security scanning before installing skills.
-
Memory Tool Upgrade — Atomic batch operations (
operationsarray). Add + replace + delete in one call, auto-balanced within token budget. Faster and more reliable behind the scenes. - New Models — GLM-5.2 (1M context), Claude Fable 5, Nemotron 3 Ultra, Laguna M.1.
- Dashboard Login Hardening — Safer if you expose your dashboard to the public internet.
- Chronos Hosted Cron — Scale-to-zero scheduled tasks. No always-on process needed.
- Japanese & Traditional Chinese UI — Desktop app now supports these languages natively.
Who Should Update?
| Your Setup | Update? |
|---|---|
| Casual chat + light coding | v0.16.0 is fine, no rush |
| Desktop app user | Strong yes — VS Code themes, notifications, sub-agent monitor are worth it |
| Frequently delegates long tasks | Strong yes — background sub-agents are a quality-of-life game changer |
| Wants scheduled tasks without cron | Strong yes — Automation Blueprints were made for you |
| iMessage / WhatsApp user | Strong yes — new integrations are notably smoother |
| Paid API user | Strong yes — Curator won't silently burn tokens anymore |
| Installed Hermes last week | Just run hermes update — one command, zero risk |
One command to update:
hermes update
That's it. The desktop app auto-restarts. All your config and memories are preserved.
v0.17.0 doesn't introduce one killer feature. It introduces a dozen small ones that collectively change when and where Hermes helps you. The direction is clear: AI agents shouldn't be confined to a chat window. They should be everywhere you are.
Top comments (0)