codex-cli
25Skill Claude CodeCodex
Delegate tasks to the locally installed Codex CLI (codex exec). Supports reasoning tiers (low to ultra), native image generation, vision input (-i), structured JSON output, and sandboxed execution.
Skill Claude CodeCodex
Delegate tasks to the locally installed Codex CLI (codex exec). Supports reasoning tiers (low to ultra), native image generation, vision input (-i), structured JSON output, and sandboxed execution.
Skill Claude CodeCodex
Execution mode. Trigger on "go ahead", "do it", "execute", "implement", "make the changes", or any approval to act. Stop planning, start building. Also governs how long-running commands, scripts, and AI CLI tools (opencode, claude -p, codex, aider, gemini, etc.) are run during execution, so that nothing can hang…
Skill Claude CodeCodex
Send clean, human-readable Slack alerts via incoming webhook. Trigger on "notify on slack", "slack alert", "ping me when done", or when a job finishes and Slack notification was requested.
Skill Claude CodeCodex
Put macOS to sleep once a task or workflow is complete. Trigger on "sleep when done", "sleep after finishing", "sleep once done", "pmset sleepnow", or requests to put the machine to sleep after task completion.
Skill Claude CodeCodex
Delegates closed-ended decisions to the Levanto Sage decision model (150ms) instead of deciding yourself. Use for command safety gates, tool routing, proceed-vs-escalate, classification, and visual pass/fail checks of Blender or three.js renders and screenshots (floating objects, black frames, missing textures…
Skill Claude CodeCodex
Delegate tasks to OpenCode Zen API (https://opencode.ai/zen/v1) and OpenCode CLI (opencode run). Supports multi-protocol auto-routing across OpenCode's top models (Fast: deepseek-v4.1-flash, Standard: kimi-k3, Deep: gemini-4-argon, qwen3.8-max. Fable is opt-in only when explicitly requested).
Skill Claude CodeCodex
Run code on Google Colab CPU, GPU (T4, L4, A100, H100), and TPU (v5e1, v6e1) runtimes via colab.
Skill Claude CodeCodex
Package an ML model (diffusion, video, LLM, any PyTorch/HF model) with Replicate Cog and push it to Replicate, from a Windows machine through WSL2 — toolchain setup (Docker Engine, nvidia-container-toolkit, cog), cog.yaml and predict.py patterns, large weights with pget, gated Hugging Face weights with per-caller…
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: