Plan Mode - Setup Guide¶
Source: nousresearch/hermes-agent (309 installs) Category: Development / Workflow License: MIT · Platforms: Linux, macOS, Windows Dependencies: None
Plan Mode switches Hermes into a planning-only execution mode. When active, the agent generates concrete, actionable markdown plans but does NOT implement code, run mutating commands, or perform external actions. Plans are saved to .hermes/plans/ in the active workspace.
What It Does¶
| Capability | How |
|---|---|
| Plan-only mode | Agent generates plans but does not execute |
| Structured output | Plans saved as markdown in .hermes/plans/ |
| Read-only inspection | Agent can inspect repos/context but not modify |
| Actionable plans | Concrete steps, not vague descriptions |
| Workspace integration | Plans live alongside project code for review |
Installation¶
Via skills.sh (Recommended)¶
npx skills add https://github.com/nousresearch/hermes-agent --skill plan
Direct from Hermes Agent Repo¶
git clone --depth 1 https://github.com/nousresearch/hermes-agent.git /tmp/hermes-agent
cp -r /tmp/hermes-agent/skills/development/plan ~/.hermes/skills/
Behavior Rules¶
When Plan Mode is active:
| Action | Allowed? |
|---|---|
| Read files / inspect repos | ✅ Yes |
| Read-only terminal commands | ✅ Yes |
| Implement code | ❌ No |
| Edit project files | ❌ No (except plan file) |
| Run mutating commands | ❌ No |
| Commit / push | ❌ No |
| External API calls | ❌ No |
Usage¶
Activating Plan Mode¶
Hermes, plan the refactoring of the authentication module
Hermes, create a plan for adding WebSocket support
Hermes, I need a plan - not execution - for migrating to PostgreSQL
Plan Output Location¶
Plans are saved to:
.hermes/plans/<timestamp>-<plan-name>.md
Plan Template¶
Generated plans follow this structure:
# [Plan Title]
## Overview
[One-paragraph summary of the goal]
## Prerequisites
- [Dependency/requirement check]
## Steps
1. [Concrete, actionable step]
2. [Concrete, actionable step]
...
## Risks
- [Identified risk] → [Mitigation]
## Estimated Effort
[Time/complexity estimate]
## Success Criteria
- [Measurable outcome]
Prerequisites¶
| Requirement | Details |
|---|---|
| Hermes Agent | v0.20.0+ |
| Workspace | Must be inside a project directory (for .hermes/plans/ output) |
Verification¶
After install, test with:
Hermes, plan how to add dark mode support to a React application.
The agent should respond with a structured markdown plan but should NOT create any files or run any code.
Pitfalls¶
- Not a gate: Plan Mode is a skill, not a system-level guard. The agent can still execute if the skill is not explicitly invoked.
- Plan quality varies: The skill provides structure but plan quality depends on the agent's understanding of the domain.
- Workspace required: Plans save to
.hermes/plans/which requires a project directory context. Outside a workspace, plans are output inline only. - One turn only: Plan mode applies to the current turn. Future turns may execute unless plan mode is re-invoked.
Installed via: npx skills add nousresearch/hermes-agent --skill plan
Integration with CorpusIQ¶
CorpusIQ is the connector layer that lets AI assistants read the systems a business already runs on - accounting, payments, CRM, email, files and analytics - and answer with the source named. It is read-only by design: it retrieves data and never writes to, or acts in, a connected system, which is what makes it safe to point at live accounts.
This page is part of the CorpusIQ directory, the reference we maintain for the MCP server and agent-skill ecosystem. Run alongside CorpusIQ, a third-party server like this one can be combined with a company's own business data in a single answer.