HagiCode is a proposal-driven AI coding workbench that turns a request into a reviewable proposal, runs it through agent CLIs such as Claude Code, Codex and GitHub Copilot, and keeps the outcome as team knowledge.
Hermes Agent is Nous Research's open-source agent with memory, skills, scheduling and chat gateways, and HagiCode is a workbench that runs Hermes as a coding executor over ACP. Use Hermes alone as an always-on assistant; add HagiCode for parallel coding sessions, proposals and commits.
Hermes Agent is an open-source, self-improving agent from Nous Research. You run hermes in a terminal, or reach it from Telegram, Discord, Slack and other chat apps, and it keeps memory and builds skills over time. HagiCode is a desktop and web AI programming workbench that drives agent CLIs, Hermes among them, as its executor. It does not replace Hermes; it runs hermes acp inside a repository and adds sessions, proposals and commit handling.
Hermes is broader than a coding agent, and that matters here. If you want an always-on assistant, scheduled jobs or a bot in your chat apps, stay with Hermes alone: HagiCode uses its ACP mode, where Hermes' own docs say messaging delivery and cron management are excluded. If your goal is coding work across repositories, HagiCode is the layer that organises it.
Open-source self-improving agent with memory, skills, scheduler and chat gateway; coding is one use
Workbench that runs agent CLIs, Hermes included, as coding executors
Terminal UI; Telegram, Discord, Slack, WhatsApp, Signal, Email; Linux, macOS, WSL2, native Windows, Termux; desktop app on macOS (Apple Silicon) and Windows
Desktop app (Windows, macOS, Linux), Docker Compose container, Microsoft Store edition
MIT-licensed and free; you pay your model provider, or an optional Nous Portal subscription
Desktop and Container free; Store edition free with optional one-time TurboEngine purchase
Nous Portal (300+ models), OpenRouter, OpenAI, your own endpoint and more; switch with hermes model
Hosts no models; uses what Hermes is configured with; optional OmniRoute routing
Spawns isolated sub-agents; seven terminal backends including Docker, SSH, Modal and Daytona
Several sessions at once, each a named Agent with visible status
No proposal workflow; learns reusable skills from experience
OpenSpec proposals: propose, validate, apply, archive (3 concurrent free, 32 with TurboEngine)
Works through its terminal tools; the README documents no dedicated commit workflow
AI-composed commit messages with the repository locked during the commit; Git and GitHub integration
One-line installer that builds its own Python environment, then hermes setup
Desktop install, plus Hermes installed with its ACP support and a provider configured
An always-on personal or team agent across terminal, chat and cron jobs
Coding work across repositories with parallel sessions and traceable specs
Hermes creates skills after complex tasks, improves them as it uses them, searches its own past conversations with full-text search and LLM summaries, and builds a model of you across sessions. It follows the agentskills.io open standard, so skills are portable. This is the feature to choose Hermes for.
One gateway process connects Telegram, Discord, Slack, WhatsApp, Signal and the CLI, with voice-memo transcription. A built-in cron scheduler takes natural-language jobs, such as nightly backups or weekly audits, and delivers results to any connected platform.
Hermes supports seven terminal backends: local, Docker, SSH, Singularity, Modal, Daytona and Vercel Sandbox. Daytona and Modal can hibernate when idle, so you can run it on a small VPS and talk to it from chat while it works on a cloud machine.
hermes model switches provider and model without code changes; the Nous Portal subscription is optional and bundles models with web search, image generation and a cloud browser. Hermes can spawn isolated sub-agents and write scripts that call its tools over RPC. Releases are frequent and still 0.x (v0.22 in October 2026).
hermes model and optionally use Nous Portal.HagiCode is a workbench on top of the CLI. It starts hermes acp in the repository you open, so your own Hermes install and provider credentials do the work. Hermes keeps reading its provider setup from its own configuration, ~/.hermes/.env or config.yaml.
Nothing is locked in: Hermes stays a normal install, with its memory, skills and gateway untouched.
curl -fsSL https://hermes-agent.nousresearch.com/install.sh | bash; Windows PowerShell: iex (irm https://hermes-agent.nousresearch.com/install.ps1)).hermes setup or hermes model to choose a provider, then check ACP with hermes acp --check; the ACP support may need its extra installed, as the Hermes ACP docs describe.hermes directly; nothing in your Hermes setup changed.Hermes can spawn sub-agents inside one conversation. HagiCode adds the level above: several Hermes sessions across tasks or repositories, each a named Agent whose status you can see.
Hermes learns from experience, but its memory is not a record of why a change was made. An OpenSpec proposal is: the problem, the agreed approach and the resulting code changes stay together and can be archived.
HagiCode drafts the commit message from the diff and locks the repository during the commit. Code Server opens the files Hermes changed in a browser editor, so review and edits stay in one window.
Monospecs declares sub-repositories in .hagicode/monospecs.yaml so a cross-repo change reaches Hermes as one proposal. Vault holds reference projects and notes shared across projects, next to what Hermes remembers about you.
OmniRoute can centralise model routing across agents. Hermes tool calls appear in the same typed view as every other agent's, and optional TokenMonitor shows today's tokens and cost in the status bar.
Hermes Agent is MIT-licensed and free to install. You pay your model provider, or optionally a Nous Portal subscription that bundles models and tools; see portal.nousresearch.com for current plans.
On Linux, macOS or WSL2 run `curl -fsSL https://hermes-agent.nousresearch.com/install.sh | bash`. On native Windows run `iex (irm https://hermes-agent.nousresearch.com/install.ps1)` in PowerShell, then start it with `hermes`.
Yes. Install Hermes and configure a provider, then choose Hermes as the Agent CLI in HagiCode. HagiCode runs `hermes acp` in your repository.
No. Hermes does the work; HagiCode organises sessions, proposals and commits around it. Its chat gateway and scheduler are not available inside HagiCode sessions.
It can code, but it is a general agent first, with memory, skills, scheduled jobs and chat gateways. HagiCode uses its coding side.
Pick Hermes alone for an always-on, multi-platform assistant. Add HagiCode when the job is coding across repositories and you want parallel sessions, proposals and managed commits.
Localized AI code assistant, privacy protection, enhanced efficiency
desktopHero.loading