aevatar-vision-probe 是做什麼的？
端對端驗證讀取圖片（視覺）路徑 — 使用者發送圖片與觸發器，工作流程的視覺步驟描述圖片內容，執行輸出證明模型確實看到了圖片（輸出實際描述，而非「未收到圖片」）。在任何依賴圖片的工作流程（如零用金收據、發票 OCR）之前執行。
端對端驗證讀取圖片（視覺）路徑 — 使用者發送圖片與觸發器，工作流程的視覺步驟描述圖片內容，執行輸出證明模型確實看到了圖片（輸出實際描述，而非「未收到圖片」）。在任何依賴圖片的工作流程（如零用金收據、發票 OCR）之前執行。
ChronoAIProject/aevatar-vision-probe 是一個專門用於驗證 AI 代理視覺管線完整性的技能。其主要目的是確保當使用者上傳圖片並觸發工作流程時，工作流程中的視覺步驟能夠正確地描述圖片內容，而不是回傳「未收到圖片」這類錯誤訊息。這項技能對於任何依賴圖片輸入的工作流程（如零用金收據處理、發票 OCR）至關重要，建議在執行這些工作流程前先執行此驗證。此技能可用於支援視覺功能的 AI 代理平台（如 Claude Code、Cursor 等），透過發送測試圖片與觸發器，並檢查工作流程的輸出是否包含實際的圖片描述，從而快速診斷視覺路徑是否正常運作。
npx skills add https://github.com/ChronoAIProject/nyx-skills/tree/main/aevatar-vision-probe端對端驗證讀取圖片（視覺）路徑 — 使用者發送圖片與觸發器，工作流程的視覺步驟描述圖片內容，執行輸出證明模型確實看到了圖片（輸出實際描述，而非「未收到圖片」）。在任何依賴圖片的工作流程（如零用金收據、發票 OCR）之前執行。
This repo contains 9 individual skills — each has its own dedicated page.
Boundary test for inbound Lark attachments — the user sends an image or file alongside this command and the bot reports exactly what it can observe on the chat path (text only, attachment refs, or nothing), without guessing. Detects when the platform wires chat-path attachment visibility so dependent skills can be upgraded.
Use when answering mathematical, conceptual, philosophical, or metaphysical questions through repository-first scientific reasoning, a clause-complete formal bridge, retained exact Lean specializations, reusable library growth, and an ordinary conversational answer; the formal work stays internal and is shown only when the user asks for it.
Send, reply to, react to, read, and find Lark messages and chats through the bot's NyxID-brokered im/v1 APIs — proactively post to a chat or DM, thread-reply, add or remove emoji reactions, read a single message or pull a chat's history over a time window, search or list the chats the bot is in, and download an image/file from a message. Handles the "content must be a JSON string" footgun and receive_id_type selection, and doubles as the im-scope probe — on a scope 403 it reports the exact missing permission instead of failing silently.
Verify-first promotion intelligence for ANY repo or org. Pulls live GitHub state, runs sshx multi-perspective consensus over it, and returns an evidence-backed board of what is genuinely promotable, how items connect, and what is unverifiable. Use when an agent or operator asks "what's worth promoting / what connects / what's real" for a target repo or org.
Method for producing short vertical videos from ChatGPT image generation through a fixed browser tab, local TTS narration, timed subtitle burn-in, cover exports, and a manifest. Use when an operator needs a reproducible image-to-short workflow with pinned session consistency, per-beat visual prompts, 1080x1920 safe-area rules, local ffmpeg/PIL assembly, and private access kept outside the skill.
The manual an AI agent loads to operate Ornn — the model-agnostic skill-lifecycle API (an npm-style registry + CLI for agent skills) — via the NyxID CLI (`nyxid proxy request ornn-api …`). Load and follow this skill WHENEVER the user asks to do anything with Ornn skills or skillsets. Skills: search Ornn or find a skill, pull or install a skill (or a specific version), run a skill, build and upload a skill, publish a new version, make a skill public / private / shared, run or read a security audit, deprecate or delete a version, diff two versions, check usage analytics, bind a skill to a NyxID service, link a skill to GitHub or sync from source, manage npm-style dist-tags, or transfer skill ownership. Skillsets — curated multi-skill bundles with a required master prompt: bundle skills into a set, create or publish a skillset, resolve its closure in one call, export a skillset as a Claude Code marketplace plugin, transfer skillset ownership, or diagnose why a shared skillset isn't visible (visibility derives fr
Runs the petty-cash approval flow end to end yourself — collect Lark DM/sheet/image context, extract receipt details, build the exact approval body, then SUBMIT the Lark approval instance.
Portable checklist for gating public-facing copy before release. Use when captions, posts, manifests, topics, titles, or launch copy must be checked for generic AI wording, leaked internal labels, forbidden experiment or tool names, platform topic allowlists, CTA spam, and unsupported overclaims.
Severity-weighted BLOCK, WARN, OK reviewer method for technical and research posts. Use when reviewing X threads, announcements, captions, or launch copy for hook strength, thread arc, generic AI vocabulary, source fidelity, claim support, numeric-claim verification, and thread format before recommending publish.
ElevenLabs text-to-speech with mac-style say UX.
Oracle CLI second-model review/debug/refactor/design with selected files, dry-run token checks, API or browser engine.
Capture and automate macOS UI with the Peekaboo CLI.
Generate multiple radically different interface designs for a module using parallel sub-agents. Use when user wants to design an API, explore interface options, compare module shapes, or mentions "design it twice".
Build a throwaway prototype to flesh out a design — a runnable terminal app for state/business-logic questions, or several radically different UI variations toggleable from one route.
Interview the user relentlessly about a plan or design until reaching shared understanding, resolving each branch of the decision tree. Use when user wants to stress-test a plan, get grilled on their design, or mentions "grill me".