Use case / editor workflow
Use Hermes Agent Inside Your Editor with ACP
Agent Client Protocol lets an editor drive Hermes Agent as a server. The practical result is that chat messages, tool activity, file diffs, and terminal commands appear in the editor you already have open, while the agent keeps the same memory and skills it uses in your messaging channels. You are not adopting a second assistant with a second memory.
- 01
Confirm the runtime already works
Run the agent once from the terminal and complete a normal exchange. Debugging the agent and the editor at the same time is how an afternoon disappears.
- 02
Check editor support
Use an ACP-compatible editor such as VS Code, Zed, or a JetBrains IDE. Support level varies, so start with the editor you use daily rather than a new one.
- 03
Register Hermes Agent as an ACP server
Point the editor at the Hermes Agent ACP entry point. The agent runs as a server that the editor starts and supervises.
- 04
Ask for one file change and read the diff
Request a small edit and inspect the rendered diff before accepting it. This verifies both the tool surface and how much review you get.
- 05
Write the review rule
Decide in advance which changes the agent may apply directly and which always need a human read. The diff view makes this enforceable rather than aspirational.
Four-step path
How do you use Hermes Agent inside an editor?
Confirm the runtime works from a terminal first, then let the editor attach to the same agent.
- Step 01
Confirm the runtime already works
Run the agent once from the terminal and complete a normal exchange. Debugging the agent and the editor at the same time is how an afternoon disappears.
- Step 02
Check editor support
Use an ACP-compatible editor such as VS Code, Zed, or a JetBrains IDE. Support level varies, so start with the editor you use daily rather than a new one.
- Step 03
Register Hermes Agent as an ACP server
Point the editor at the Hermes Agent ACP entry point. The agent runs as a server that the editor starts and supervises.
- Step 04
Ask for one file change and read the diff
Request a small edit and inspect the rendered diff before accepting it. This verifies both the tool surface and how much review you get.
- Step 05
Write the review rule
Decide in advance which changes the agent may apply directly and which always need a human read. The diff view makes this enforceable rather than aspirational.
Why run the agent inside the editor?
Because the work happens next to the files, so diffs and terminal commands are visible as they are proposed instead of arriving as text in a chat window.
- File changes arrive as reviewable diffs.
- Tool activity is visible while it happens rather than summarised afterwards.
- The same runtime, memory, and skills serve editor and chat.
- No second subscription or second configuration surface to maintain.
Which editors are supported?
ACP-compatible editors including VS Code, Zed, and JetBrains. The agent speaks the protocol, so editor choice stays a preference rather than a migration.
Because the integration is protocol-based, an editor that gains ACP support works with the same configuration. There is nothing Hermes-specific to port.
- VS Code for the widest extension ecosystem.
- Zed for a lighter editor with fast startup.
- JetBrains IDEs where the team is already invested.
How should a team review agent changes?
Treat agent edits exactly like a colleague's pull request: small scope, a visible diff, and a person who owns the merge.
- Keep one task per request so the diff stays readable.
- Never accept a diff that touches secrets or deployment configuration unreviewed.
- Re-run the tests the change affects before merging.
- Record which tasks the agent handled so the practice can be audited.
What are the limits of editor integration?
The editor is a view onto the agent, not a replacement for a terminal, and it does not change what the agent is allowed to touch on the host.
- Permissions still come from the runtime, not from the editor.
- A long autonomous task belongs in a scheduled run, not in a foreground editor session.
- Very large repositories still need the agent pointed at the right subtree.
What pairs well with editor integration?
Tool servers for the systems you look things up in, and a shared channel so the same agent answers questions outside your editor.
Which primary sources support this page?
Product behavior is checked against the current deployment flow. These external links provide the upstream project or channel documentation used for setup details.
Direct answers
Frequently asked questions
Is editor integration a separate product?
No. It is the same runtime exposed over ACP, so the agent keeps the memory, skills, and tool servers you already configured.
Does the agent need a server to work in an editor?
Not necessarily. It can run locally for editor work, and move to a server or managed deployment when other people need to reach it.
Can I use the editor and a messaging channel at once?
Yes. Both talk to the same runtime, so a question asked in chat and a change requested in the editor see the same context.
Key takeaways
- Agent Client Protocol lets an editor drive Hermes Agent as a server. The practical result is that chat messages, tool activity, file diffs, and terminal commands appear in the editor you already have open, while the agent keeps the same memory and skills it uses in your messaging channels. You are not adopting a second assistant with a second memory.
- Start with one model, one channel, and one controlled test conversation.
- Use managed hosting when deployment speed matters more than operating the server yourself.
Continue the setup
Related Hermes Agent guides
Use case
Connect MCP Tool Servers to Hermes Agent
Connect Hermes Agent to MCP tool servers for GitHub, databases, file systems, note apps, and internal APIs without writing native tools.
Read nextUse case
Hermes Agent Integrations for Team Tools
Connect Hermes Agent to the tools your team already uses, including Google Workspace, calendars, note apps such as Notion or Joplin, and MCP servers.
Read nextBlog guide
Hermes Agent Configuration Guide: Models, Providers, and Fallbacks
Configure Hermes Agent models, providers, fallback chains, and runtime defaults, and know which settings you can change in a managed deployment.
Read nextUse case
Hermes Agent Web Dashboard for Hosted Deployments
Manage a hosted Hermes Agent from a web dashboard: check runtime health, review sessions, restart a deployment, and control channels without SSH or server access.
Read next