24 skills shared by ChatGPT, Claude Code, and Codex.

Each skill tells the agent what it does and when to use it, so the right workflow can activate without you memorizing its name.

See what each skill does and when it steps in →

Installation

The repository contains the 24-skill 2.2.0 candidate. The official Plugins Directory remains the recommended distribution for ChatGPT and Codex, but its published version can lag this source tree. Claude Code can install the current source catalog from the repository marketplace.

ChatGPT

Open the Plugins Directory in ChatGPT, search for Skillquiver, note the displayed version, and select Add. ChatGPT uses @ mentions when you want to select a bundled skill explicitly. The new 24-skill catalog is available there only after the listing shows version 2.2.0.

Claude Code

Inside an interactive Claude Code session:

/plugin marketplace add Drizzy07x/Skillquiver
/plugin install skillquiver@skillquiver

Or from a terminal:

claude plugin marketplace add Drizzy07x/Skillquiver
claude plugin install skillquiver@skillquiver --scope user

Claude Code installs all 24 skills. Update later with:

/plugin update skillquiver@skillquiver

or:

claude plugin update skillquiver@skillquiver

For a manual global install, clone the repository and copy the shared catalog:

git clone https://github.com/Drizzy07x/Skillquiver.git
mkdir -p ~/.claude/skills
cp -r Skillquiver/skills/* ~/.claude/skills/

Use a project's .claude/skills/ directory instead for a project-only install.

Codex

From a terminal:

codex plugin marketplace add Drizzy07x/Skillquiver
codex plugin add skillquiver@skillquiver

Codex installs all 24 skills. Refresh the marketplace and reinstall the plugin to update:

codex plugin marketplace upgrade skillquiver
codex plugin add skillquiver@skillquiver

For a manual global install:

git clone https://github.com/Drizzy07x/Skillquiver.git
mkdir -p ~/.codex/skills
cp -r Skillquiver/skills/* ~/.codex/skills/

Use a project's .agents/skills/ directory instead for a project-only install. Plugins are supported in Codex CLI and the Codex desktop app; use the manual project install for clients that do not load plugins.

Important

v2.2 candidate: improve-agent-instructions adds scoped AGENTS.md and CLAUDE.md maintenance. Skillquiver Doctor moved into skills/ in v2.1. Existing marketplace installs update through their host commands above. Manual installations copy only skills/ for the complete 24-skill catalog.

Repository layout

Skillquiver/
├── skills/                              # 24 skills shared by all three hosts
├── .claude-plugin/plugin.json           # exposes all 24 to Claude Code
└── .codex-plugin/plugin.json            # exposes all 24 to Codex

Skills

Shared compatibility badge:

Planning

Skill	Compatibility	What it does
brainstorming		Explores user intent, requirements, and design before implementation.
writing-plans		Writes a task-by-task implementation plan document from a spec, before touching code.
executing-plans		Executes a written implementation plan inline, task by task, stopping only on blockers.
subagent-driven-development		Executes a plan by dispatching one subagent per task with per-task review gates, keeping orchestrator context small.
dispatching-parallel-agents		Dispatches parallel subagents, one per independent domain, to handle unrelated problems concurrently.

Execution

Skill	Compatibility	What it does
test-driven-development		Enforces test-first red-green-refactor: failing test, minimal code to pass, verify green.
solve-efficiently		Routes work efficiently with progressive context discovery, matched effort, and durable project mapping.
handle-host-boundaries		Handles unavailable host-specific skills and tools without fabricating capabilities or crossing host boundaries.
execute-durably		Runs long or interruption-prone work against an external state file with falsifiable criteria and an append-only evidence log.
refactor-safely		Refactors working code and lands changes in untested legacy code without altering observable behavior.
using-git-worktrees		Ensures an isolated workspace exists via native tools or git worktree fallback before feature work.
communicate-clearly		Controls report verbosity with the shortest profile that preserves evidence; explains external sources in plain language.

Verification & debugging

Skill	Compatibility	What it does
diagnose-systematically		Finds the cause of a defect through observable evidence: reproduces, minimizes, tests falsifiable hypotheses one variable at a time.
verification-before-completion		Requires running verification commands and confirming output before any success claim; evidence before assertions.
verify-work		Independently audits finished work: verifies a delivery or completion claim against real evidence.
research-systematically		Freezes the question before collecting results, binds every claim to a source, pins docs to the installed dependency version.

Code review

Skill	Compatibility	What it does
requesting-code-review		Dispatches a code-reviewer subagent to check finished work against requirements before it merges.
receiving-code-review		Processes review feedback with rigor: verify each claim against the code before implementing, push back with evidence.
finishing-a-development-branch		Guides integration of a finished branch: verify tests, present merge/PR/keep/discard options, clean up worktrees.

UI & automation

Skill	Compatibility	What it does
design-ui		Turns visual intent into inspectable constraints, commits to a stated visual direction, builds it as one system, and verifies the rendered result.
automate-ui		Drives web and desktop UIs adaptively while capturing evidence that separates navigation from behavioral proof.

System

Skill	Compatibility	What it does
improve-agent-instructions		Audits and improves global, project, and nested AGENTS.md and CLAUDE.md guidance while preserving host-specific behavior.
skillquiver-doctor		Audits host-local skills, plugins, and hooks for demonstrated conflicts, then offers reversible repairs with per-item consent.

Prompt engineering

Skill	Compatibility	What it does
engineer-prompts		Builds or audits a prompt contract with explicit outcomes, boundaries, permissions, required evidence, and stop conditions.

Support and legal

• Publisher: Drizzy07x
• Support
• Privacy policy
• Terms of use
• MIT license