aidev TIER 1 harness Nous Research

Hermes Agent

RELEASES24
BREAKING5
LATESTv2026.8.3
SHIPPED8d
TRACKED SINCE2026-03-12
FULL TIMELINE → OFFICIAL SITE ↗ § MARKDOWN

RELEASE CADENCE · LAST 12 MONTHS

09
10
11
12
01
02
03
04
05
06
07
08

BREAKING CHANGES

v2026.5.29 Hermes Agent v0.15.1: dashboard reload-loop hotfix, Docker --insecure now explicit 2mo v2026.5.28 Hermes Agent v0.15.0: agent-loop refactor, Kanban orchestration, rebuilt session_search 2mo v2026.5.7 Hermes Agent v0.13.0: durable multi-agent Kanban, session auto-resume, 8 P0 security fixes 3mo

RECENT

8d
16:57
aidev hermes-agent

Hermes Agent v0.20.0: streaming voice with barge-in, outbound webhooks, desktop plugin SDK

Large release centred on real-time conversational voice: Hermes now speaks clause-by-clause as it generates, can be interrupted mid-sentence, and responds to on-device wake words across CLI, desktop, and messaging gateways. It also gains signed outbound webhooks for pushing lifecycle events to external HTTP endpoints, a grounded-citations research skill, and a desktop plugin SDK with sandboxed artifact previews.

featurecapabilityapi-change
11d
23:45
22d
18:35
aidev hermes-agent

Hermes Agent v0.19.0: ~80% faster first token, smart approvals on by default, password-manager secrets

Large Hermes release cutting cold-start time to first token from ~4.3s to ~0.9s across CLI, gateway, TUI, desktop and cron, with reasoning now streamed live by default. It also flips flagged-command approvals to LLM-reviewed by default, adds Bitwarden/1Password as secret sources, live subagent transcripts, and a durable ledger so a finished response survives a gateway crash. Anyone running Hermes unattended or with approval gates should read the approval-behaviour change before upgrading.

featurecapabilitysecurity
1mo
03:11
hermes-agent Hermes Agent v0.18.2: WhatsApp Baileys dependency unpinned v2026.7.7.2 · SIG 2/5
1mo
01:15
hermes-agent Hermes Agent v0.18.1: untriaged rollup tag of ~667 commits v2026.7.7 · SIG 2/5
1mo
20:08
aidev hermes-agent

Hermes Agent 0.18.0: Mixture-of-Agents as a selectable model, evidence-based completion

Major release. Mixture-of-Agents stops being a mode and becomes a selectable model under a moa provider in every picker, with each reference model's output shown as its own block and the aggregator's answer streamed live. The agent now records verification evidence for coding work and judges completion against /goal completion contracts rather than self-assertion, with a pre_verify hook and a one-time defaults migration. Also adds /learn to distil reusable skills and /journey plus a desktop memory graph to inspect and prune what the agent has learned.

featurecapabilitymodel-support
1mo
19:39
aidev hermes-agent

Hermes Agent v0.17.0: iMessage and Raft channels, background subagents, image editing

Large release adding two new places Hermes can run — iMessage via Photon's managed line pool (no Mac relay or BlueBubbles bridge) and the Raft agent network via a wake-channel bridge — plus background subagent delegation, image-to-image editing in image_generate, and grok-composer-2.5-fast reachable through an xAI Grok OAuth subscription. The desktop app and dashboard both gained substantial surface. Matters if you run Hermes for yourself or a team on any of these channels.

featurecapabilitymodel-supportsecurity
2mo
00:55
aidev hermes-agent

Hermes Agent v0.16.0: native desktop app and browser admin panel

Hermes ships an Electron desktop application for macOS, Linux and Windows with in-app self-update, and the web dashboard becomes a full administration surface for MCP, channels, credentials and auth. The desktop client can also drive a remote Hermes gateway over WebSocket with OAuth or username/password, so the GUI and the agent runtime no longer have to live on the same machine. The bundled skill set was trimmed, with several skills removed outright and others demoted to optional installs.

featurecapabilitysecuritydeprecation
2mo
13:37
hermes-agent Hermes Agent v0.15.2: bundled plugin manifests included in packages v2026.5.29.2 · SIG 2/5
2mo
01:12
aidev hermes-agent BREAKING

Hermes Agent v0.15.1: dashboard reload-loop hotfix, Docker --insecure now explicit

Same-day hotfix for v0.15.0. The headline fix stops the dashboard full-page reload loop that hit every loopback-mode install (Docker, hosted, fresh installs), where a by-design 401 from /api/auth/me was mistaken for a rotated session token. Docker users who relied on --insecure being inferred from a non-loopback bind host must now set HERMES_DASHBOARD_INSECURE=1 explicitly — that is a required action, not a default carry-over.

fixbreakingsecurity
2mo
17:46
aidev hermes-agent BREAKING

Hermes Agent v0.15.0: agent-loop refactor, Kanban orchestration, rebuilt session_search

The conversation loop in run_agent.py drops from 16,083 to 3,821 lines across 14 agent/* modules with forwarders kept for compatibility, Kanban becomes a multi-agent orchestration platform, and session_search is rewritten without an aux-LLM (free, ~20ms discovery instead of ~90s and ~$0.30/call). Anyone scripting Kanban, relying on the session_search config knob or companion skill, or running Grok models retired on May 15 should read before upgrading.

featurecapabilitysecuritybreakingdeprecationapi-changemodel-support
2mo
09:59
aidev hermes-agent

Hermes Agent v0.14.0: SuperGrok OAuth, OpenAI-compatible local proxy, PyPI install

Hermes ships on PyPI, adds xAI Grok through SuperGrok OAuth with grok-4.3 raised to a 1M-token context window, and exposes hermes proxy — an OpenAI-compatible localhost endpoint backed by whichever OAuth subscription you are signed into, so Codex CLI, Aider, Cline and Continue can run on it without an API key. Heavy backends now lazy-install on first use, cutting install size, and cold start drops ~19 seconds.

featurecapabilitymodel-supportcontext-windowapi-change
ALL 24 RELEASES →