self-customize

Modify agent capabilities by installing packages, adding MCP servers, and delegating code changes to builder agents.

2|Updated Jan 14, 2026
One-click install
npx skills add https://github.com/RADHA0-max/selfhealbackend --skill self-customize-radha0-max
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: self-customize
Source: https://github.com/RADHA0-max/selfhealbackend/tree/main/nanoclaw/container/skills/self-customize
Command: npx skills add https://github.com/RADHA0-max/selfhealbackend --skill self-customize-radha0-max

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Agents often need new capabilities—system tools, MCP servers, or source code changes—but making these changes ad hoc risks unreviewable edits and broken environments. This Skill provides a structured decision tree and safe workflows for an agent to modify its own environment with proper approval boundaries. ## Core Features & Use Cases - Decision Tree Routing: Directs each change type (workspace files, apt/npm packages, MCP servers, source code) to the correct workflow with the right approval level. - Builder Agent Delegation: Spins up a dedicated builder agent via create_agent for source code edits, enforcing diff size limits (200 new / 150 modified lines) so changes stay reviewable. - Package & MCP Installation: Uses install_packages and add_mcp_server with admin approval, triggering automatic image rebuilds or container restarts. - Use Case: A user asks the agent to transcribe audio. The agent checks for ffmpeg, requests install_packages with apt and npm packages, waits for admin approval, and tests the new capability after the automatic container restart. ## Quick Start Ask the agent to add a new capability, such as "add a tool for reading RSS feeds", and it will route the request through the appropriate self-customization workflow.

Frequently Asked Questions about self-customize

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I add a new MCP server to my agent?

Use the add_mcp_server function with the server name, command, and arguments. This requires admin approval, and on approval the container restarts with the new server wired up, with no image rebuild needed since bun runs TypeScript directly.

How do I install system packages in an agent container?

Call install_packages with the apt or npm packages and a reason. Admin approval is required, and once approved the container image is rebuilt and the container restarts automatically with the new packages available.

Why delegate code changes to a builder agent instead of editing directly?

Delegating to a builder agent creates a reviewable boundary and keeps the main session focused. The builder enforces diff limits of 200 new lines or 150 modified lines per task, forcing large changes into smaller reviewable chunks.

Can I edit CLAUDE.md directly to change agent behavior?

No, the composed CLAUDE.md is read-only and regenerated on every spawn. Write persistent customizations to CLAUDE.local.md in the workspace instead, which is persisted on the host and requires no approval.

When should I not use self-customization workflows?

Avoid self-customization for one-off tasks that can be done in the workspace, ambiguous requests that need clarification first, or unproven approaches. Prototype in the workspace with pnpm install before promoting to a container-level install.