菜单
Use when the input is an image, screenshot, diagram, PDF/DOCX, audio file, or YouTube link—extract metadata/text before reasoning.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Use when shell commands fail, npx/curl/git won't run, or you must choose run_shell vs web_fetch vs file tools in Nodebox vs host.
Use when the user says remember this, save a preference or API key, or you must pick memory_save vs session notes vs skill (action=manage) vs wiki_* tools.
Use when the user wants to discover, search, or install agent skills from online registries—top results by installs, stars, or votes. Or user sends /find_skills.
Use when calling REST or GraphQL APIs with web_fetch/web_post — auth headers, query params, GraphQL shape, Directus GraphQL mutations/relations, discovery order, and reading error bodies.
Use when the user or agent installs a skill from skills.sh or GitHub — map WebFetch, Bash, Python, Playwright, and MCP references to Web Agent built-ins.
Use when starting a new app, demo, spike, starter, or multi-file project—make_dir under projects/ or work/ before any writes.
| name | Multimodal Ingest |
| description | Use when the input is an image, screenshot, diagram, PDF/DOCX, audio file, or YouTube link—extract metadata/text before reasoning. |
| version | 1.1.0 |
| category | bundled |
| primary-tools | ["vision_analyze","image_info","audio_analyze","youtube_transcribe","pdf_extract","docx_extract"] |
| tags | ["vision","multimodal","ocr","youtube","transcript","image","pdf","docx"] |
| triggers | ["screenshot","image","diagram","ocr","what is in this image","youtube link","transcribe","video summary","pdf text","docx text","vision_analyze","youtube_transcribe","audio_analyze"] |
| Input | Tool |
|---|---|
| Image metadata / dimensions | image_info |
| Image / screenshot / diagram | vision_analyze with a focused question |
| YouTube / Shorts URL | youtube_transcribe |
| Workspace audio file | audio_analyze |
| Workspace PDF | pdf_extract |
| Workspace DOCX | docx_extract |
| Cross-check claims | web_search, web_fetch — open-web-research |
| Persist transcript / extract | write_file under work/<slug>/ |
| Show visual to user | artifact_present — artifact-delivery |
| Durable facts | memory_save — memory-layers |
Non-negotiable: Focused vision questions, not "describe everything". Long transcripts → file + present, not chat paste.
open-web-research. Present files: artifact-delivery. Facts: memory-layers.vision_analyze with a focused question, not "describe everything". Examples: "What error code is shown?", "List visible CLI flags", "Read the chart x-axis values".youtube_transcribe → grep / scan segments for the relevant span → summarize with timestamp citations ([mm:ss]). For long videos, transcribe once, then operate on the cached transcript text.pdf_extract / docx_extract before run_python; persist long extracted text under work/<slug>/.open-web-research (verify a name, date, or product before asserting).memory_save; long transcripts belong in work/<slug>/ then artifact-delivery — not inline.vision_analyze on the same image for variant questions — combine questions or cache the prior answer in session memory.