milady-development

The Milady Development skill empowers your AI agent with the ability to perform self-directed improvements and development tasks. Designed for users who want to extend their agent beyond its base capabilities, this skill provides a structured interface to interact with the underlying plugin architecture. By utilizing this skill, the agent can navigate the local file system, identify plugin entry points, and apply modifications to actions, providers, or services without impacting the stable elizaOS core. It is intended for power users, developers, and autonomous agents that need to evolve their behavior, integrate custom logic, or optimize their responses over time.

• Perform hot-reloads and full runtime restarts via RESTART_AGENT actions or local API endpoints to apply new code immediately.

• Manage the entire lifecycle of plugins: install from registry, eject for local modification, sync with upstream changes, and create new custom plugins from scratch.

• Integrate custom TypeScript logic into plugins to define new actions, context providers, or background services that influence the LLM reasoning process.

• Utilize a safe-by-default architecture that explicitly protects the read-only elizaOS core infrastructure while granting freedom within the user-editable ~/.milady/plugins/ directory.

• Leverage command-line tools to monitor plugin health, update repositories, and verify dependencies using package managers like bun and npm.

• Expected inputs include natural language commands for plugin management (e.g., 'Eject plugin-twitter') or file-path references for code modifications.

• The skill assumes access to a local development environment where node.js, bun, and git are installed to support the build and merge process.

• Practical constraints: Always treat the core runtime (~/.milady/eliza) as immutable. Modifying the core can result in breaking changes during future upstream updates.

• For best results, use the RESTART_AGENT action after modifying any TypeScript files in the plugins folder to ensure the registry reloads the new module exports correctly.

• Use this skill to implement specific agent improvements, such as custom Discord/Telegram logic, advanced data providers, or proprietary evaluators that fit your specific privacy and automation needs.