Hermes Agent: CLI Command Line
Last updated: 2026-08-31
The CLI is Hermes Agent's most powerful operation mode — all features can be accessed via command line, supporting script integration, batch operations, and automation pipelines.
💡 Tip: The CLI is the most efficient way for developers to use Hermes. Desktop and Web versions cover 80% of features, CLI covers 100%. Automation scenarios require the CLI.
📋 Prerequisites: Lesson 2 Installation, Lesson 3 Configuration File
1. What You Will Learn
| # | Content |
|---|---|
| ❶ | CLI command overview |
| ❷ | Core command details |
| ❸ | Conversation and interaction commands |
| ❹ | Management and operations commands |
| ❺ | Script integration and automation |
2. Story
(1) Pain Point: Manual Operations Every Time
Bob's daily routine: open Hermes → switch model → start conversation → save results → send notification. 5 manual steps, tedious and easy to forget.
(2) Solution: One Command Does It All
Alice wrote a script that runs automatically every day:
BASH
#!/bin/bash
# daily-workflow.sh
hermes chat --model gpt-4o --skill daily-briefing --output report.md
hermes notify telegram --file report.md
hermes checkpoint save "daily-$(date +%Y%m%d)"
3. CLI Command Overview
(1) Command Categories
| Category | Command | Description |
|---|---|---|
| Core | hermes chat |
Interactive conversation |
hermes serve |
Start API server | |
hermes run |
Single task execution | |
| Config | hermes init |
Initialize |
hermes config |
Configuration management | |
hermes model |
Model management | |
| Memory | hermes memory |
Memory management |
| Skills | hermes skill |
Skill management |
| Tools | hermes tool |
Tool management |
| Platforms | hermes platform |
Platform management |
| Scheduled | hermes cron |
Scheduled tasks |
| Checkpoints | hermes checkpoint |
Checkpoint management |
| Delegation | hermes delegate |
Task delegation |
| Plugins | hermes plugin |
Plugin management |
| Dashboard | hermes dashboard |
Data dashboard |
| Profile | hermes profile |
User configuration |
4. Core Command Details
(1) hermes chat
BASH
# Basic conversation
hermes chat
# Specify model
hermes chat --model gpt-4o
# Specify user
hermes chat --user alice
# Specify skill
hermes chat --skill code-review
# Non-interactive mode (single input)
hermes chat --message "Review src/main.py" --no-interactive
# Pipe input
cat error.log | hermes chat --message "Analyze this error log"
# Specify toolset
hermes chat --toolset frontend-dev
# Output to file
hermes chat --output report.md
(2) hermes run
BASH
# Single task execution
hermes run "Search React 19 new features and generate summary"
# Specify model and skill
hermes run --model gpt-4o --skill research "Research Vue 4 changes"
# Delegation mode
hermes run --delegate --max-sub-agents 3 "Review code quality of entire project"
# Output format
hermes run --format json "Analyze this API's response format"
(3) hermes serve
BASH
# Start API server
hermes serve --host 0.0.0.0 --port 8080
# Run as daemon
hermes serve --daemon
# Specify configuration
hermes serve --config /path/to/config.yaml
# Enable all platforms
hermes serve --platforms telegram,discord,slack
5. Conversation and Interaction Commands
(1) In-Conversation Commands
In hermes chat interactive mode, commands starting with /:
| Command | Function |
|---|---|
/learn <content> |
Teach Agent a new skill |
/remember <content> |
Save memory |
/forget <id> |
Delete memory |
/skill <name> |
Invoke skill |
/model <name> |
Switch model |
/toolset <name> |
Switch toolset |
/profile <name> |
Switch user |
/checkpoint save <name> |
Save checkpoint |
/checkpoint restore <name> |
Restore checkpoint |
/delegate status |
View delegation status |
/honcho profile |
View user model |
/cost |
View current session cost |
/help |
Help |
/exit |
Exit |
(2) Conversation Keyboard Shortcuts
| Shortcut | Function |
|---|---|
↑/↓ |
Browse input history |
Tab |
Auto-complete command/skill/tool names |
Ctrl+C |
Interrupt current operation |
Ctrl+D |
Exit conversation |
Ctrl+L |
Clear screen |
6. Management and Operations Commands
(1) Configuration Management
BASH
# View configuration
hermes config show
hermes config get model.default
hermes config get memory.working_memory_limit
# Set configuration
hermes config set model.default gpt-4o
hermes config set personality.tone professional
# Validate configuration
hermes config validate
# Export/Import
hermes config export > backup.yaml
hermes config import backup.yaml
# Reset
hermes config reset
(2) Model Management
BASH
# List available models
hermes model list
# Test model connection
hermes model test gpt-4o
# Add custom model
hermes model add --name my-model --provider ollama
# View model details
hermes model info gpt-4o
(3) Health Check
BASH
# Full health check
hermes doctor
# Quick check
hermes health
# View version
hermes --version
# View system info
hermes info
7. Script Integration and Automation
(1) Shell Script Integration
BASH
#!/bin/bash
# CI/CD integration example
# 1. Code review
hermes run --skill code-review --output review.md "Review PR #$PR_NUMBER"
# 2. Check review results
if grep -q "🔴 Critical" review.md; then
echo "Critical issues found, blocking merge"
exit 1
fi
# 3. Run tests
hermes run --skill test-runner "Run full test suite"
# 4. Generate release notes
hermes run --model gpt-4o --output CHANGELOG.md "Generate release notes from git log"
(2) Python Integration
PYTHON
from hermes import HermesClient
client = HermesClient(base_url="http://localhost:8080")
# Send message
response = client.chat(
message="Review the code quality of this file",
model="gpt-4o",
files=["src/main.py"]
)
print(response.text)
# Batch execution
results = client.batch_run([
{"message": "Search React 19", "model": "gpt-4o-mini"},
{"message": "Search Vue 4", "model": "gpt-4o-mini"},
{"message": "Search Svelte 5", "model": "gpt-4o-mini"},
])
for r in results:
print(f"Task: {r.task}, Cost: ${r.cost:.4f}")
(3) Makefile Integration
MAKEFILE
.PHONY: review test deploy
review:
hermes run --skill code-review "Review all modified files"
test:
hermes run --skill test-runner "Run test suite"
deploy: review test
hermes run "Deploy to staging environment"
@echo "Deployment complete"
❓ FAQ
Q Does CLI support Windows?
A Yes. Both PowerShell and CMD work. Some pipe features work better in WSL2.
Q How to view command help?
A
hermes --help for global help, hermes chat --help for subcommand help.Q What is non-interactive mode suitable for?
A Script integration, CI/CD, scheduled tasks, batch processing — any automation scenario without human interaction.
Q How to run multiple tasks in parallel?
A Shell background
hermes run ... & or Python multithreading. CLI itself supports --delegate for parallelism.Q Where are CLI logs?
A Default
~/.hermes/logs/. hermes chat --verbose shows detailed real-time logs.Q How to debug CLI issues?
A
hermes chat --debug enables debug mode with full tool calls and model requests output.📖 Summary
- CLI covers 100% of features, the most efficient operation mode for developers
hermes chatfor interactive conversation,hermes runfor single tasks,hermes servefor API server- In-conversation
/commands: skills, memory, model, checkpoints - Management commands: config, model, health, doctor
- Script integration: Shell, Python, Makefile for automation pipelines
📝 Exercises
- Basic (⭐): Use CLI for a complete workflow: initialize → chat → save checkpoint → view cost.
- Intermediate (⭐⭐): Write a Shell script using
hermes runto automate code review + testing + report generation. - Advanced (⭐⭐⭐): Develop an automation workflow with Python HermesClient: daily scheduled Git log fetch → generate daily report → send notification, integrated with cron.
