install mise
Installing Mise and Setting Up the Environment
To install the mise polyglot development tool and environment manager, run the setup command specific to your operating system.
Step 1: Download and Install the CLI
The recommended way to install on macOS is via Homebrew Formulae:
brew install mise
Alternatively, you can download a standalone binary using the mise-en-place interactive script:
curl https://mise.run | sh
Note that this script installs the binary to ~/.local/bin by default.
For Windows, install mise using Scoop or Winget:
winget install jdx.mise
Step 2: Activate Mise in Your Shell (Recommended)
To allow mise to automatically switch programming language versions and load environment variables when you change directories, add its hook to your shell configuration.
For Bash, run:
echo 'eval "$(~/.local/bin/mise activate bash)"' >> ~/.bashrc
For Zsh, run:
echo 'eval "$(~/.local/bin/mise activate zsh)"' >> ~/.zshrc
For Fish, run:
echo '~/.local/bin/mise activate fish | source' >> ~/.config/fish/config.fish
If you installed via Homebrew or a package manager, replace ~/.local/bin/mise with just mise in the snippet above.
Step 3: Verify and Start Using
Verify that everything is set up correctly by running the built-in troubleshooter:
mise doctor
Test out installing a tool (like Node.js) globally:
mise use --global node@26
If you prefer not to activate the full shell hook, you can skip step 2 entirely and execute commands isolated inside a temporary environment using mise exec node@26 -- node -v.
Troubleshooting Oh-My-Zsh Issues
When encountering issues with Oh-My-Zsh, such as the error zsh: command not found: pygmentize, it's essential to identify the responsible plugin.
The Oh-My-Zsh plugins responsible for this error are common-aliases or colorize. These plugins fail when they attempt to use pygmentize, a Python-based syntax highlighter that is not installed on your system or missing from your $PATH.
Fixing the Issue
You have three options to fix this issue:
- Install Pygments: Install Python and add the global
pygmentspackage.mise use --global python@latest pip install pygments - Remove the Problem Plugins: Open your configuration file, find the
plugins=(...)line, and removecommon-aliasesand/orcolorizefrom the list. Save and restart your terminal. - Unalias the Capital "P": Disable the specific alias by appending
unalias P 2>/dev/nullto the bottom of your~/.zshrcfile.
Resolving Ollama Errors
When encountering errors with Ollama, such as the "pulling manifest Error: launching hermes: failed to pull gemma4: pull model manifest: open /Users/avery/.ollama/id_ed25519: no such file or directory" error, it's essential to understand the cause.
This error occurs because Ollama's internal cryptographic key pairs are either missing, corrupted, or deleted from your computer.
Fixing the Ollama Error
To fix this issue, completely shut down Ollama and let it regenerate its missing security keys automatically.
- Quit Ollama: Click the Ollama icon in your Mac's top menu bar and click "Quit Ollama."
- Force-Stop the Process: Run
killall Ollamato ensure the process is stopped. - Remove the Directory: Run
rm -rf /Users/avery/.ollama/id_ed25519*to remove the directory. - Restart Ollama: Open your Mac's Applications folder and double-click the Ollama app to start it back up.
When Ollama boots up and detects that id_ed25519 is missing, it will silently generate a brand new, valid public/private key pair in /Users/avery/.ollama/ right away.
Configuring Hermes-Agent
To configure Hermes-Agent, you need to create a config.yaml file with the correct structure.
Unified Hermes-Agent Configuration
Here is an example of a unified config.yaml file:
# ==============================================================================
# UNIFIED HERMES-AGENT CONFIGURATION
# ==============================================================================
# 1. Primary Orchestrator Settings
model: hermes3:8b
provider: ollama
api_base: http://localhost:11434/v1
temperature: 0.2
max_tokens: 4096
# 2. Enabled Core Toolsets
toolsets:
- hermes-cli
- web
# 3. Sub-Agent Delegation Registry
delegation:
enabled: true
fallback_to_orchestrator: true
# Specialized Coding Sub-Agent
code_expert:
model: qwen2.5-coder:7b
provider: ollama
api_base: http://localhost:11434/v1
temperature: 0.1
system_prompt_override: >
You are a specialized code execution and repository scanner sub-agent.
Analyze the subfolders carefully. Return ONLY functional code or raw
directory maps. Do not truncate JSON blocks.
# Fast Content Summarization Sub-Agent
summary_helper:
model: gemma4
provider: ollama
api_base: http://localhost:11434/v1
temperature: 0.4
system_prompt_override: >
You are a lightning-fast content summarization agent. Extract key bullet
points from text files or video transcripts provided by the orchestrator.
This configuration defines a primary orchestrator with a delegation layout using local models.
Verification Check
To verify that your configuration is correct, check the following:
- Model Tags: The
qwen2.5-coder:7btag matches Ollama's default naming exactly. - Gemma 4 Tag: Ensure
gemma4is the exact name returned when you runollama list. If it was pulled asgemma4:latest, you can safely leave it asgemma4or explicitly typegemma4:latestin themodel:line.
By following these steps and verifying your configuration, you should be able to set up and use Hermes-Agent with Ollama successfully.