---
name: agentic-coding-harnesses
description: Use when modifying the agentic coding environment, including dotfiles, skills/rules wiring, AGENTS.md/CLAUDE.md loading, Stow symlinks, Claude/Codex/OpenCode/Pi config, plugins, MCP, hooks, or harness compatibility. Loads the runbook to avoid breaking cross-harness setup. Defer to skill-creator-global for skill authoring craft, description quality, and progressive disclosure.
---

# Agentic Coding Harnesses Reference

Discovery skill that points to the comprehensive runbook for the user's multi-harness agentic environment.

## Read the runbook first

Before making any change to skills, rules, AGENTS.md, plugin configuration, MCP setup, or harness wiring, read:

```
~/.dotfiles/docs/AGENTIC-CODING-HARNESSES.md
```

The runbook is the source of truth for the canonical layout and procedures.

## What the runbook covers

- Per-harness loading mechanics for skills, rules, and instruction files (Claude Code, Codex CLI, OpenCode, Pi)
- Canonical file layout in `~/.dotfiles/`
- Procedures for: adding a skill, modifying imported skills, trimming AGENTS.md, handling plugin re-install conflicts, re-stowing, debugging skill discovery failures
- Size caps and adherence guidance per harness (Codex 32 KiB hard cap, Claude < 200 lines, "lost in the middle")
- Provenance tracking (user-authored vs. Matt Pocock vs. plugin vs. Claude.ai-app)
- Decision log explaining the architecture

## When to invoke this skill

Whenever the user's prompt involves modifying their agentic environment. Following the runbook prevents breaking the cross-harness setup. If a procedure isn't documented yet, add it to section 16 of the runbook before executing the change.