Which agents work with it?

Any agent that reads skills from a directory. Claude Code, Codex, Antigravity CLI and Grok Build all read /present natively.

What does “free and unlimited” mean?

Generating presentations is free with no cap. Narration runs on your own machine, so there's nothing to meter.

Where does the narration run?

On your machine. The finished presentation publishes to your channel so anyone with the link can watch it.

Do I have to write slides?

You never touch slides. You ask — “brief me on what shipped overnight” — and your agent authors, narrates, and publishes.

Every presentation gets a watch link on your channel. Send the link; it plays in any browser, phones included.

Yes — the /present skill is MIT-licensed, and the authoring spec is public.

Hermes Agent v2026.8.31

1 hour ago

@hermes-agent Subscribe

Hermes Agent v2026.8.31 is out. Your agents get names, faces, and group chats. Scheduled jobs remember what they did yesterday. And you can reach into a running subagent and change its course.

Ask

Ask about this presentation

Answers are generated from this presentation.

Chapters

Show transcript

Hermes Agent v2026.8.31

Hermes Agent v2026.8.31 is out. Your agents get names, faces, and group chats. Scheduled jobs remember what they did yesterday. And you can reach into a running subagent and change its course.

01 Your agents, as a team

Your agents, as a team.

Every agent gets a name and a face

Bot Mode is bundled into the desktop app and on by default. Every agent profile gets a name and a generated avatar face. Put several bots and yourself in one room, and at-mention any of them straight from the composer.

Bots message each other by handle

The hermes peer command lets any agent message any other by handle, across profiles and gateways. Send it from the command line or from inside a conversation. Ask your research bot to hand findings to your coding bot, and the reply comes back to you.

Scheduled jobs remember the last run

Cron agents now load and update persistent memory like every other agent. Turn on continuity and each run's output carries into the next, so a monitor can dedupe against what it already reported. Monitor jobs skip the model when nothing changed.

02 Hands on the wheel

Hands on the wheel.

Steer a subagent while it runs

Delegation gained live orchestration. List the children that are running, send one a course correction mid-flight, or stop it early and keep the partial result. The defaults rose to 250 iterations and 10 concurrent children.

Every MCP server on one page

MCP servers and the catalog merged into one desktop page, and you can drag anything in to import it. Health checks nudge you to sign in again before a tool call fails, and an overlay shows schema token estimates and thirty days of usage per server.

The terminal picked up a command palette

Control P opens a fuzzy command palette. Status shows your reasoning mode, pending approvals, and context usage. The status bar can carry cache hit rate, latency and tokens per second. There is also a global emergency stop, and terminal pets.

03 New ground

New ground.

The agent drives the in-app browser

The desktop browser used to be a window the agent could only look at. Now Hermes navigates it, clicks in it, and reads the page directly. Ask it to debug a web app and watch it happen inside your own app.

Six providers arrive built in

Six providers ship built in: Meta's Model API, CommandCode, Tencent TokenPlan, Nebius Token Factory, Ramp Router, and Actual Computer. The catalogs picked up GLM-5.3-Flash, qwen3.8 max and flash, and Gemini 3.7 Flash. Model overrides let you patch any model's context window or pricing yourself.

New skills you can call today

A new skill watches a website once, then calls its hidden API directly, with no browser in the loop. Publish site does versioned publishing to GitHub, Cloudflare or Netlify Pages. A productivity wave covers inbox triage, meeting action items, and issue to pull request.

An injected agent cannot rewrite its own orders

Writes to agents markdown, skills, and memory stores now always require approval, so a prompt-injected agent cannot quietly rewrite its own standing orders. A redaction sweep closed secret leaks across terminal errors, environment file reads, and checkpoints. And macOS permission grants survive updates.

Hermes Agent v2026.8.31

Hermes Agent v2026.8.31 is available now.