Hermes Agent v0.21.0 (v2026.8.31)¶
Release Date: August 31, 2026 Since v0.20.0: ~5,800 commits · ~2,475 merged PRs · ~5,680 files changed · ~2,100 issues closed · 760+ contributors
The Pantheon Release. v0.20.0 made Hermes the herald. In v0.21.0 the gods assemble: a society of named agents that talk to each other like a team, cron jobs that remember between runs, subagents you can steer mid-flight, an MCP command center, and an agent that drives the desktop's own browser. This release also fully documents the v0.20.1 through v0.20.6 infrastructure patch windows.
Highlights¶
- Bot Mode, built into the desktop app - every agent profile gets a name, a deterministic avatar face, and a place in a shared roster. Create group chats where multiple bots and you talk in one room, @-mention any bot from the composer, and give rooms names and pictures. Multi-agent now looks like a chat app full of coworkers.
hermes peer- bot-to-bot DMs - any Hermes agent can message any other by handle, across profiles and gateways, from the CLI or from inside a conversation.- Cron jobs that remember - scheduled jobs load and update persistent memory like any other agent.
continuity=truecarries each run's output into the next run, so scheduled agents learn between ticks. - Steer your subagents while they run -
delegate_taskgained live orchestration: list running children, steer one mid-flight with a course correction, or stop it early and keep the partial results. - The MCP command center - MCP servers and the catalog merged into one coherent desktop page with drag-in import, background health checks that nudge re-auth before a tool fails, and one-click connects.
- A CLI power wave - Ctrl+P fuzzy command palette,
/modelpicker that filters as you type,/statusshowing reasoning mode, pending approvals, and context usage. - The agent drives the desktop's browser - Hermes now navigates, clicks, and reads the in-app browser directly instead of only observing it.
- Six new providers - Meta Model API (Muse Spark) as a built-in provider, alongside CommandCode, Tencent TokenPlan, Nebius Token Factory, Ramp Router, and Actual Cloud.
- Security hardening - protected agent-instruction files (AGENTS.md, skills, memory stores) now always require write approval, so a prompt-injected agent cannot quietly rewrite its own rules.
Updating¶
hermes update
# or fresh install:
curl -fsSL https://hermes-agent.nousresearch.com/install.sh | bash
Full Changelog: v2026.8.3...v2026.8.31
← v0.20.1 - Patch Release | v0.21.1 - Patch Release →
This Hermes repo is one of the largest structured collections of public AI, automation, business, and technology documentation. Content remains attributed to original authors and repositories. Indexed and organized by www.CorpusIQ.io.