codex-workflow-skills

Use when Codex needs to ask DeepSeek v4 Pro a focused one-to-one question through a local Python wrapper instead of MCP. Default to DeepSeek for second-model consultation unless the user explicitly names Codex, codex exec, or the Codex provider; only then use the isolated Codex CLI backend.

Locate, inspect, and safely summarize local Codex conversation logs by UUID. Use when a user provides a Codex conversation/thread UUID and asks to read, explain, recover, continue, hand off, 查看、接續、恢復、回看、理解 prior Codex Desktop/CLI thread context without exposing raw private transcripts.

Broad local Codex session search and safe activity summarization with distill. Use when the user asks what they did today, yesterday, recently, this week, or during a time range; asks to summarize Codex work history without providing a UUID; says things like "看看昨天做什麼", "今天做了什麼", "最近我在忙什麼", "整理一下這幾天的 Codex 對話"; or wants a dominant workstream summary across many local Codex Desktop/CLI sessions. For exact UUID recovery, hand off to codex-conversation-lookup instead.

Use when debugging bugs, incidents, flaky tests, regressions, production issues, or unclear system behavior where the root cause is not yet bounded. Applies when Codex should increase observability first by adding targeted logs, metrics, traces, or assertions before proposing a fix. Triggers include requests to debug, investigate, instrument, add logs, narrow down, reproduce, or "wrap the problem with logs".

Use when creating or revising Xiaohongshu/XHS share image cards in Figma for any project. Trigger when the user asks to turn a Blog, markdown draft, product idea, launch note, or existing Figma canvas into Xiaohongshu image-card designs. Use Figma as the review and assembly surface; route generated bitmap visuals, poster-style covers, illustrations, and style-heavy cards through imagegen before placing the selected output back into Figma.

Use when preparing, validating, filling, or recording a Xiaohongshu/XHS post after the user has approved the content and images. This skill handles the publishing handoff: validate title, body, hashtags, image order, and image dimensions; optionally open the Xiaohongshu creator composer through the user's existing Arc CDP session; optionally upload images and fill title/body; stop before the final publish click unless the user explicitly delegates that live action.