From ai-workspace
Initialize the ai-workspace plugin in the current directory. Creates threads/, AGENTS.md, CLAUDE.md, and .claude/settings.json. Safe to re-run, skips files that already exist.
How this skill is triggered — by the user, by Claude, or both
Slash command
/ai-workspace:initThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Initialize the ai-workspace plugin in the current working directory.
Initialize the ai-workspace plugin in the current working directory.
threads/ — directory where conversation threads will be storedAGENTS.md — workspace instructions read by both Claude Code and Codex CLICLAUDE.md — one-line pointer to AGENTS.md (Claude Code reads this natively).claude/settings.json — permission allowlist for plugin operations on Claude CodeAll steps are skipped if the file or directory already exists. Safe to re-run.
The same files work for both Claude Code and Codex CLI. Codex ignores CLAUDE.md and .claude/; Claude ignores files it doesn't recognize. No vendor detection needed.
threads/mkdir -p threads
Skip if threads/ already exists.
AGENTS.mdCheck if AGENTS.md exists. If not, call mcp__plugin_ai-workspace_threads__get_skill_file("templates/AGENTS.md.template") and write the result to AGENTS.md.
CLAUDE.mdCheck if CLAUDE.md exists. If not, call mcp__plugin_ai-workspace_threads__get_skill_file("templates/CLAUDE.md.template") and write the result to CLAUDE.md. The template is a single @AGENTS.md line so Claude imports the AGENTS.md content without content duplication.
.claude/settings.jsonCheck if .claude/settings.json exists. If not:
mkdir -p .claude
Then call mcp__plugin_ai-workspace_threads__get_skill_file("templates/settings.json.template") and write the result to .claude/settings.json.
This file is Claude-Code-specific. It is harmless on Codex; Codex ignores it.
This step only applies on Codex CLI. Detect the CLI before doing anything: you know which CLI you are. If you are Claude Code, skip this entire step — Claude loads the proponent and skeptic agents directly from the plugin's agents/ directory; nothing to copy.
If you are Codex CLI: Codex plugin manifests cannot bundle subagents directly, so the .toml files have to be copied into the user's home agents/ directory:
mkdir -p ~/.codex/agents
cp <plugin-root>/.codex-plugin/agents/*.toml ~/.codex/agents/
Resolve <plugin-root> from the location of this SKILL.md (three directories up: .. → .. → ..). Skip files that already exist in ~/.codex/agents/ unless the user asks to overwrite.
Report what was created vs skipped:
Initialized ai-workspace in /path/to/workspace
Created:
threads/
AGENTS.md
CLAUDE.md
.claude/settings.json
Skipped (already exists):
(none)
Start your first thread:
/ai-workspace:threads create my-first-thread (Claude Code)
/threads create my-first-thread (Codex CLI)
On Claude Code, restart the CLI for the new permissions to take effect.
Use the actual absolute path of the current working directory, not a placeholder.
If .claude/settings.json was newly created, remind the user to restart Claude Code so the permission allowlist takes effect. This restart is not needed on Codex.
npx claudepluginhub sebmartin/ai-marketplace --plugin ai-workspaceProvides UI/UX resources: 50+ styles, color palettes, font pairings, guidelines, charts for web/mobile across React, Next.js, Vue, Svelte, Tailwind, React Native, Flutter. Aids planning, building, reviewing interfaces.
Fetches up-to-date documentation from Context7 for libraries and frameworks like React, Next.js, Prisma. Use for setup questions, API references, and code examples.