Skills are prompt-based instructions injected into an agent's system prompt to guide decisions, workflows, and tool use. Unlike MCP servers, they run no process — each skill is a focused playbook an agent activates when the task calls for it. Oh My Opencode Slim bundles 8 skills, installed and kept up to date by the installer.
Repository Cartography
Generates hierarchical codemap.md files that document the why and how of each directory, so agents understand unfamiliar codebases without re-reading thousands of lines of code on every task.
codemap Full docs 
Heavy Session Workflow
A workflow for large, high-risk, multi-phase coding efforts with meaningful dependencies and review gates. Reserved for the sessions that are too big to wing.
deepwork Full docs 
Evidence Before Code
Designs a credible, project-specific evidence path before non-trivial implementation work — features, bug fixes, refactors, and cross-system changes get a verification plan first.
verification-planning Full docs 
Behavior-Preserving Refactors
Simplifies code for clarity without changing behavior. Focused on readability, maintainability, and complexity reduction once the existing behavior is understood.
simplify Full docs 
Isolated Coding Lanes
Manages Git worktrees as safe, isolated coding lanes — so complex, risky, or parallel work never contaminates your main checkout.
worktrees Full docs 
Dependency X-Ray
Clones important dependency source code into an ignored local workspace, letting agents inspect library and SDK internals instead of guessing from documentation.
clonedeps Full docs 
Workflow Self-Improvement
Reviews recent work for repeated patterns and suggests reusable skills, agents, commands, config changes, or playbooks — turning recurring friction into permanent improvements.
reflect Full docs 
Plugin Self-Configuration
Configures and safely improves the plugin itself — tuning agents, models, prompts, custom agents, skills, MCPs, and presets for the current user.
oh-my-opencode-slim Full docs The Orchestrator holds most skills by default; review-oriented skills belong to Oracle. Assignments are permission grants — an agent can only activate skills it has been given.
| Skill | Purpose | Assigned to |
|---|---|---|
| codemap | Repository Cartography | @orchestrator |
| deepwork | Heavy Session Workflow | @orchestrator |
| verification-planning | Evidence Before Code | @orchestrator |
| simplify | Behavior-Preserving Refactors | @oracle |
| worktrees | Isolated Coding Lanes | @orchestrator |
| clonedeps | Dependency X-Ray | @orchestrator |
| reflect | Workflow Self-Improvement | @orchestrator |
| oh-my-opencode-slim | Plugin Self-Configuration | @orchestrator |
Yes. Skill assignments are configurable per agent in ~/.config/opencode/oh-my-opencode-slim.json — grant an explicit skill list, allow everything with *, or deny individual skills with a !skill-name entry.
By Alvin