heretic-cli

heretic-cli

Command-line interface for managing containerized AI coding agents with layered configuration, secure secrets, and multi-provider support

CLI-First Approach

All agent configuration is done through CLI commands. The CLI automatically creates and manages:

  • Secret scripts — Shell scripts that output API tokens securely
  • Claude settings — Claude Code permissions and model configuration
  • Agent profiles — Full agent configuration with provider details

Never edit config files manually

Use CLI commands for all changes to ensure consistent configuration and proper secret handling.

Key Features

  • Multi-provider support — Anthropic, third-party APIs (ZAI, Kimi), Copilot, and Kilocode
  • Layered configuration — Global profiles + per-project overrides + CLI flags
  • Secure secrets — Script-based token management, never stored in YAML
  • MCP integration — Model Context Protocol server configuration with format auto-detection
  • Sessions — Run multiple isolated agent instances in the same project
  • Cross-platform — Linux, macOS, and Windows support

Quick Overview

# Set up global settings and create an agent profile
heretic-cli init

# Create per-project configuration
heretic-cli local-init claude

# Run the agent
heretic-cli run claude

Next Steps