ponytail
121humanerd-drew/opencode-drewgent
Skill Claude CodeCodex
Forces AI agents through a minimization checklist before writing code. 80-94% less code, 3-6x faster. Adapted from DietrichGebert/ponytail.
Drewgent configuration for opencode — agent profiles, skills, scripts, and tooling
This repository also configures its own agents. See what opencode-drewgent tells them →
humanerd-drew/opencode-drewgent
Skill Claude CodeCodex
Forces AI agents through a minimization checklist before writing code. 80-94% less code, 3-6x faster. Adapted from DietrichGebert/ponytail.
humanerd-drew/opencode-drewgent
Skill Claude CodeCodex
A payment integration guide for PortOne, a Korean payment gateway service that connects web apps to providers such as KG Inicis, Kakao Pay, and Toss Payments. It covers browser-side payment requests and server-side verification.
humanerd-drew/opencode-drewgent
Skill Claude CodeCodex
Systematic methodology for analyzing a project's file structure, mapping dependency graphs, identifying dead/orphan code, and removing it safely with local verification.
humanerd-drew/opencode-drewgent
Skill Claude CodeCodex
How to patch large Python files in the {{AGENTNAME}} codebase. Covers the dual-file layout (runtime /.{{AGENTNAMELOWER}}/agent/ vs source /.{{AGENTNAMELOWER}}/source/{{AGENTNAMELOWER}}-agent/), diagnose-fallback when direct patch fails, and the copy-sync protocol to keep both copies aligned.
humanerd-drew/opencode-drewgent
Skill Claude CodeCodex
An explanation of a Python scoping bug caused by importing a module inside a function or nested function. Python then treats that name as local within the surrounding function, which can cause an error before the import runs.
humanerd-drew/opencode-drewgent
Skill Claude CodeCodex
Rename {{AGENTNAME}} to your own agent name — replaces all references across 2000+ files, renames directories, updates configs.
humanerd-drew/opencode-drewgent
Skill Claude CodeCodex
Code review workflows: pre-commit diff verification and full project codebase audit.
humanerd-drew/opencode-drewgent
Skill Claude CodeCodex
Token-efficient default for searchfiles tool — adds includecontent and previewchars params so matches are truncated by default (86% token savings on long lines).
humanerd-drew/opencode-drewgent
Skill Claude CodeCodex
Audit and fix shell init files (zsh/bash) where a side effect (bg fetch, daemon, cache) runs in a context where its consumer (RPROMPT, prompt, key binding) is gated OFF. Trigger when user shows [N] PID job notification on shell login, asks to silence one, or asks to modify .zshrc/.zshenv/.zprofile/.zlogin/.bashrc. Fix…
humanerd-drew/opencode-drewgent
Skill Claude CodeCodex
Parallel 3-agent cleanup of recent code changes.
humanerd-drew/opencode-drewgent
Skill Claude CodeCodex
Throwaway experiments to validate an idea before build.
humanerd-drew/opencode-drewgent
Skill Claude CodeCodex
Dispatch parallel subagents for implementation and refactoring — two-stage review, wave-based batch delegation, and mechanical split patterns.
humanerd-drew/opencode-drewgent
Skill Claude CodeCodex
2026-06-23: Superseded by GJC Coordinator MCP. OMO subagent profile system (delegate.ts + agents/.md) was replaced by Gajae-Code. Use task() for simple subagent work (opencode built-in) or gjcdelegate for isolated/parallel tasks.
humanerd-drew/opencode-drewgent
Skill Claude CodeCodex
Multi-phase debugging protocol — find root cause before fixing. Covers the 4-phase process, {{AGENTNAME}} stack debugging, and T4-style stall investigation.
humanerd-drew/opencode-drewgent
Skill Claude CodeCodex
Token-efficient default for terminal tool — adds taillines/headlines opt-in params and lowers default cap from 50K to 5K (40/60 head/tail split). 90% token savings on verbose commands.
humanerd-drew/opencode-drewgent
Skill Claude CodeCodex
Write the test first. Watch it fail. Write minimal code to pass.
humanerd-drew/opencode-drewgent
Skill Claude CodeCodex
Write comprehensive implementation plans assuming the implementer has zero context for the codebase and questionable taste. Document everything they need: which files to touch, complete code, testing commands, docs to check, how to verify. Give them bite-sized tasks. DRY. YAGNI. TDD. Frequent commits.
humanerd-drew/opencode-drewgent
Skill Claude CodeCodex
Patch {{AGENTNAME}} config.yaml safely — handles the dual-config pattern (/.{{AGENTNAMELOWER}}/config.yaml + P5-ego/config/config.yaml) and avoids mcppatch YAML gotchas.
humanerd-drew/opencode-drewgent
Skill Claude CodeCodex
Skill "specification-website" from humanerd-drew/opencode-drewgent, covering specification-website, when to use this skill, quickest path — mcp, without mcp — fetch markdown directly and categories.
humanerd-drew/opencode-drewgent
Skill Claude CodeCodex needs its repo
A review process for studying a selected tool and recording the design decisions worth applying elsewhere. It researches the tool's documentation, architecture, user experience, and distinctive choices.
humanerd-drew/opencode-drewgent
Skill Claude CodeCodex
A procedure for checking MiniMax coding-plan usage and reset time from a terminal or a shell status line. It uses MiniMax's usage endpoint and can connect the result to a zsh prompt.
humanerd-drew/opencode-drewgent
Skill Claude CodeCodex
Skill "baseline-ui" from humanerd-drew/opencode-drewgent, covering baseline ui, when to apply, rule categories by priority, 1. stack defaults and 2. html semantics.
humanerd-drew/opencode-drewgent
Skill Claude CodeCodex
UI/UX design agent skill. Creates design mockups, UI components, design systems, and visual assets. Uses Lazyweb for references, baseline-ui for conventions. Does NOT implement backend logic.
humanerd-drew/opencode-drewgent
Skill Claude CodeCodex
Yuanbao (元宝) groups: @mention users, query info/members.
At most 3 mods per repository are shown here, and a mod shipped inside a plugin is left to that plugin's page — the rest are on their repository pages: