Converts various media into structured notebook notes.
Install
mkdir -p .claude/skills/note-v1ceversaa && curl -L -o skill.zip "https://agentskills.codes/api/skills/download/16520" && unzip -o skill.zip -d .claude/skills/note-v1ceversaa && rm skill.zipInstalls to .claude/skills/note-v1ceversaa
Activation
This is the description your AI agent reads to decide when to run this skill — the better it matches your request, the more reliably it fires.
Convert PPTs, slide decks, PDFs, Word documents, lecture transcripts, articles, papers, screenshots, rough bullets, or existing excerpts into V1CeVersa's Notebook notes. Use this skill whenever the user asks to read material and write, expand, rewrite, or place a note in this MkDocs notebook, even if they only say "整理成笔记", "做成 notebook 风格", "读一下这个 PPT", "根据这篇文章生成笔记", or provide a target docs/ path.Key capabilities
- →Convert various source materials into MkDocs-native notes
- →Restructure material into a concept hierarchy
- →Draft notes in Chinese with technical precision
- →Review notes for text quality and image placement
- →Apply text corrections and finalize image slots
- →Integrate new notes into existing MkDocs navigation
How it works
The skill processes source material through four stages: orient, draft, review, and revise, to produce notes that adhere to the V1CeVersa Notebook style. It focuses on structuring content conceptually and ensuring linguistic and technical accuracy.
Inputs & outputs
When to use note
- →Create a note from a PDF document
- →Summarize an article into a notebook page
- →Convert PPT slides to structured notes
About this skill
V1CeVersa Notebook Note Skill
Produce notes that explain the material coherently, cover the requested knowledge completely, and use no more space than the explanation needs. Accept source files, URLs, excerpts, and an optional target path. Follow the user's requested scope and deliverable; a source or docs/ path alone does not authorize writing a note.
Reader and Stance
These are the owner's study notes. Write as the owner: a student who has worked through the material and now explains it in their own words to their future self. That reader comes back months later knowing what the earlier notes in the series cover. They want to understand the mechanism and be able to use it. They are not auditing the source, tracking version history, or looking up a reference manual. Do not write as a translator of the source, a fact-checker, or a documentation maintainer.
The stance decides where each piece of information goes:
- Body: the problem, the concept, the mechanism, the conditions that change how the reader writes code or carries out an argument, and the example that makes the mechanism visible. A version number belongs here only when it changes what the reader can write, such as a feature that needs a minimum version or a default that changed.
- Collapsible block (
???- Info): exhaustive enumerations, edge cases, implementation details, and version history. These are things the reader may want later but does not need in order to follow the main explanation. - HTML comment: how a claim was checked. This covers tested versions and environments, documentation discrepancies, source pages, and, when the note departs from the source, the source's original claim and why. The corrected rule itself goes in the body.
- Omit: details that neither help understanding nor change usage.
Completeness and source fidelity mean the reader gets every idea they need, and none of them distorted. They do not mean that every qualification found in the source or produced by a verification pass appears in the body. When a review or audit comes before a revision, its findings decide what to fix. Its evidence and phrasing stay out of the prose.
The register is written Chinese that stays easy to follow, neither chatty nor stiff. Target Register in human-voice.md defines it and gives examples of both ways to miss it.
Choose the Work Needed
Mechanical edits — typos, links, and local formatting: read the affected passage and enough context to avoid changing its meaning. Check the changed text or link directly; do not load writing guides or inspect unrelated notes.
Content patches — a definition, derivation, example, or section: read the affected section and follow its dependencies (symbols, assumptions, references, and subsequent uses). Read the whole note only when those dependencies require it. Match the target's existing conventions; inspect a nearby note only if the target leaves a convention unclear. Apply the writing guidance to new or modified prose, not to untouched paragraphs. Update an outline when section coverage changes. Do not add navigation or reorganize files unless needed for the requested deliverable.
New notes or substantial rewrites — read the source within the requested coverage, identify its conceptual dependencies, and inspect representative nearby notes. Use the references below to plan and draft. Expanding a short outline entry is a patch; completing an entire stub follows this route. An existing file can take either route according to the user's request.
Revising existing notes takes whichever route above fits its size, with five additions.
- Text the owner wrote is kept, and only its errors are fixed. That covers both the oldest committed version and the owner's own in-place edits.
- Do not re-add content the owner deleted. If a deletion looks accidental, restore it and say so. An example is a whole subsection or a cross-reference that vanished during an unrelated edit.
- Before a substantial rewrite, diff the file against its committed and staged versions. This catches content that earlier rewrites dropped.
- The user may name an earlier version as the register reference. Then keep its paragraphing and sentence register, and carry over the newer facts.
- Report restored content, and every change to sentences the owner wrote, as a list the owner can veto.
Placement and navigation changes — inspect the relevant mkdocs.yml subtree and local path conventions. A request for a new navigable topic includes the necessary index and minimal nav wiring; a single-note patch does not. Use the user's target when given. Ask only if unresolved placement would materially change the notebook's long-term structure; continue independent reading meanwhile.
Read References as Needed
- notebook-style.md: formatting and file organization. Read the relevant sections when creating a note or changing its structure; use the target's conventions for a small patch.
- human-voice.md: explanatory, connected Chinese prose, common AI tells, and verbatim excerpts from the owner's notes for calibration. Read when drafting or substantially revising prose.
- domain-patterns.md: coverage questions for a new note or substantive section. Read only the relevant domain. These are prompts for deciding what the material needs, not mandatory headings or a fixed order.
- image-workflow.md: read when acquiring, inserting, or modifying images. Skip when the task needs no image work.
Local series conventions guide formatting, not factual correctness.
Source and Coverage
For a full chapter, lecture, or paper note, read the whole requested source unit, including relevant figures, captions, tables, assumptions, and limitations. For a targeted addition, read the relevant passages and their dependencies. Do not infer unread coverage from a title or abstract. If extraction is incomplete, identify the gap and recover the affected material before claiming complete coverage.
Organize a learning note by conceptual dependencies while preserving all substantive content needed for the requested scope. Re-derive the source's points in the order the reader needs them, rather than translating its sentences or carrying over its rhetorical setups. Follow source order when the user requests faithful translation or sequential commentary. For slides, recover the connections between definitions, diagrams, and derivations; for articles, retain the argument and qualifications. Cut repeated exposition rather than necessary reasoning.
Source fidelity is always required. Distinguish reported claims, explanatory derivations, and your own evaluations. Derive added explanations from stated premises or support them with sources. Do not invent missing empirical results, citations, causal explanations, or consensus. Keep assumptions and uncertainty close to the claims they qualify.
Draft, Review, and Illustrate
Draft the conceptual explanation, checking coverage and dependencies as it develops. Review content and structure before polishing wording. Correct only the requested scope; flag unrelated problems separately. The workflow can revisit a previous decision when reading or verification reveals an error.
Inspect source figures when needed to understand the material. Usually finish text structure before bulk image processing. During drafting, a temporary <!-- FIGURE: concept; source page --> comment can mark an image still needed. Choose images for the information they contribute, with no required count or density. For several images, keep a compact working record of placement, purpose, and source to avoid repeated searches.
Acquire, process, and insert the needed assets using the image reference. If inspecting a figure reveals a factual or structural error, correct the affected text and check it again. Text is not frozen against corrections.
Completion and Verification
Finish the requested local edits and their applicable checks without asking for review between routine steps. Follow AGENTS.md for preview and build selection; do not run a repository build for prose-only edits.
A complete delivery covers the requested material, preserves qualifications, and includes the necessary figures and explanations. Check changed outline entries, links, anchors, nav paths, and image paths against their actual targets. After inserting images, inspect the resulting assets for legibility, complete labels, and agreement with the prose; inspect the rendered page when sizing or placement needs validation.
Temporary figure comments and unchecked outline entries are not substitutes for required content. Remove obsolete placeholders; if a necessary source or asset remains unavailable, state exactly what is incomplete, finish independent work, and do not describe the note as complete. An explicitly requested draft may retain clearly identified gaps.
Stop verification after relevant checks pass unless a new change, failure, or unresolved concern warrants another check. Report the deliverable, checks performed, and material limitations concisely.
When not to use it
- →When the user wants a chat summary
- →When the user wants to insert actual images during drafting
- →When the user wants to update `mkdocs.yml` nav, create `index.md`, or reorganize directories without explicit instruction
Limitations
- →Notes must avoid self-descriptive organization prose
- →Notes must not include actual images during the drafting stage
- →Notes must not contain generic transitions, empty overview paragraphs, or self-referential explanations
How it compares
This skill produces learning artifacts organized by concepts with specific quality checks, unlike a generic summarization tool that might generate a simple overview.
Compared to similar skills
note side by side with the closest alternatives in the catalog.
| Skill | Installs | Updated | Safety | Difficulty |
|---|---|---|---|---|
| note (this skill) | 0 | 5mo | No flags | Advanced |
| super-save | 1 | 5mo | Review | Beginner |
| article-add | 0 | 10mo | No flags | Beginner |
| braindump | 7 | 9mo | No flags | Beginner |
Try saying
Example prompts that trigger this skill in your AI assistant.
More by V1CeVersaa
View all by V1CeVersaa →You might also like
super-save
supermemoryai
Save important project knowledge to memory. Use when user wants to preserve architectural decisions, significant bug fixes, design patterns, or important implementation details for team reference.
article-add
petestewart
Add a topic to the article queue for later generation. Use when the user invokes "/article-add <topic>" or says "add to article queue", "queue up an article about", or "save this topic for an article later".
braindump
huytieu
Quick capture of raw thoughts with intelligent domain classification and competitive intelligence extraction
knowledge-ops
ThanhTrunggDEV
Knowledge base management, ingestion, sync, and retrieval across multiple storage layers (local files, MCP memory, vector stores, Git repos). Use when the user wants to save, organize, sync, deduplicate, or search across their knowledge systems.
note
yangyuan-zhen
[OMX] Save notes to notepad.md for compaction resilience
meeting-minutes
github
Generate concise, actionable meeting minutes for internal meetings. Includes metadata, attendees, agenda, decisions, action items (owner + due date), and follow-up steps.