agentleFS
yokingmayokingma/flow-diagram-skill/skills/flow-diagram/SKILL.md

flow-diagram

Skill1 starsChanged 3 days agoMITView on GitHub

Use when the user asks to create or revise a flowchart, architecture diagram, sequence diagram, or animated process explanation(流程图、架构图、时序图、流程动画演示). Create executable FlowSVG diagrams with static views or animation; validate or diagnose existing diagrams; export JSON, offline HTML, SVG, PNG or GIF; manage project diagram preferences.

2 other projects use a version of this.

Adopted by

Most recently taken up by

  • Other versions2 projects use a version of this.

    Word for word · 0

    None found.

    Edited · 0

    None found.

    Wrote their own version · 2

    1. tejastn10/workbench★ 1same name
    2. omnigami-ai/pro-skills★ 0same name

    How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.

  • In this repositoryOne other file it gives its agents.

The file

Show its rules and text

We split this skill into the rules it gives an agent

13 rules. About 0 are common, 11 are distinctive, and 2 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈11

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 2

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Create accurate, understandable and visually clear explanations.

    line 8
  2. Distinctive:

    Start with the reader's question: what should this diagram explain?

    line 27
  3. Distinctive:

    Before arranging nodes and edges, use diagram selection to match the reader's question and source relationships to a diagram type.

    line 31
  4. Distinctive:

    Read the branches relevant to the work:

    line 41
  5. Distinctive:

    Text and containers: use typography, spacing, color and

    line 44
  6. Distinctive:

    Display destination: read PPT conditions or article conditions for those uses.

    line 46
  7. Distinctive:

    Geometry assistance: read measurement when text sizing needs it.

    line 47
  8. Distinctive:

    Animation: read purpose, inherited state, transfer/emphasis and effective timing and

    line 48
  9. Distinctive:

    Extra formats: read SVG/PNG, named batches or GIF when requested.

    line 49
  10. Distinctive:

    Use read-only validation while editing:

    line 57
  11. Very specific:

    Read finalization. Publish current JSON, paused offline HTML, report.json, report.md and requested formats:

    line 85
  12. Very specific:

    If issues remain, summarize their effects in Agent-authored Markdown and pass --notes <summary.md>.

    line 91
  13. Distinctive:

    After interruption, inspect remaining task files and continue the unfinished work.

    line 101

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Flow Diagram

Create accurate, understandable and visually clear explanations. Design standards guide the work; judge the result by actually viewing it against the reader's question and display conditions. Invest in reading, measurement, rendering and revision when they improve the result.

This skill produces diagram assets. PPTX and native PowerPoint shapes are outside its scope. Use Node 20+ and npm; run node <skill-directory>/scripts/flowsvg.mjs from the user's chosen working directory. Resolve tools and references inside this installed skill. Its pinned SDK runtime leaves application dependencies untouched.

Choose the task

RequestAction
New or revised diagramFollow the process below; preserve imported facts and requested formats outside the authorized changes.
Continue unfinished workRead its source draft, issues.md if present and current report; resume the affected steps. See source selection.
Export without editsPreserve selected JSON bytes and page design. Read exports, then generate, view and deliver the requested files.
Validation onlyRun read-only validate, report exact checks, limits and findings, then stop.
Visual diagnosis without editsFollow inspection preparation, view the requested surfaces, report findings and gaps, then stop.
Save, change or remove a project preferenceFollow preference writing. A one-off choice stays in this task; saving requires explicit intent to reuse it.

Default process

1. Understand

Start with the reader's question: what should this diagram explain? Read the user's designated sources or existing diagram within the authorized scope. When using designated source material, record its references and scope through source metadata. Apply relevant project preferences to open choices.

Use known display conditions and requested formats. For missing conditions, adopt and state the existing defaults; known PPT use remains PPT use. A clear brief can proceed directly to creation. Clarify factual gaps that change meaning, key relationships, boundaries or scope; explain SDK limits that prevent faithful expression before choosing an alternative.

Before arranging nodes and edges, use diagram selection to match the reader's question and source relationships to a diagram type. Read the matched semantic pattern; use free composition when no listed type covers the main question.

Finish when the reader's question and supporting facts are clear, and either a primary diagram type and its applicable guidance are selected or free composition has a defined main relationship. An independent plan is optional.

2. Create

For every new or revised draft, read document model, shared quality standards, visual direction and visual inspection guidance. Reuse sections already loaded.

Organize content to answer the question. Reorder, group, merge duplicate explanations and select abstraction within the source meaning. Use source fidelity when unknowns or information reduction affect interpretation. Ask before changing facts, key relationships, boundaries or scope.

Read the branches relevant to the work:

  • Relationship scope: with groups or cross-boundary edges, choose relationship scope and endpoints.
  • Text and containers: use typography, spacing, color and cards/groups/tags. With edges, read connections. Select a composition recipe when it fits; consult only relevant shape/label/routing sections in rendering gotchas.
  • HTML: apply page content and editable pages; use the starter when useful. Preserve existing page design through page continuation.
  • Display destination: read PPT conditions or article conditions for those uses. Independent images need their own readable context.
  • Geometry assistance: read measurement when text sizing needs it. For repeated geometry, use layout workflow and layout-plan contract; irregular compositions can use direct coordinates.
  • Animation: read purpose, inherited state, transfer/emphasis and effective timing and rhythm. With captions, check their continuous reading windows. Choose stages that help explain changes or transfers.
  • Extra formats: read SVG/PNG, named batches or GIF when requested.

Before saving, read task workspace. Keep one editable source draft at the task-workspace root; update that file throughout creation. Current HTML, requested images and machine report are replaced when regenerated. Keep one issues.md for material questions, diagnostics, visible problems and failures, updating location, status and handling result. Normal steps and full stdout need no log; CLI does not parse or require this Markdown.

Finish when the draft expresses the selected content and relationships. Concrete design choices belong in the draft; separate proof of design compliance is unnecessary.

3. Validate

Use read-only validation while editing:

node <skill-directory>/scripts/flowsvg.mjs validate --input <task-workspace>/draft.json

Correct exact input/reference/SDK errors. Read diagnostics when findings occur. Geometry and text estimates are optional aids with stated methods and limits; visible readability and connection clarity belong in viewing. When needed, measure --input <draft> --in-place and layout --input <draft> --plan <plan> --in-place update the same source draft.

Generate the current result from that source:

node <skill-directory>/scripts/flowsvg.mjs author --input <task-workspace>/draft.json --name <name> --output <task-workspace>/history

Read the compact result and use its real detailsPath for detailed timing or capture facts. Default capture covers actual HTML; --preview adds SDK state captures for a specific diagnosis. --samples adjusts needed animation capture coverage; --browser none skips capture for intermediate HTML. Parameters remain discoverable through --help.

Finish when exact errors are fixed and requested inspection files exist, or an execution gap and its affected files are identified. A successful machine check is an execution fact.

4. View and revise

Read actual HTML inspection. View the saved HTML, its captured images and requested SVG/PNG at intended dimensions and background. Trace important edges to their arrowheads; assess text, hierarchy, boundaries and reading order with the design guidance. For animation, read key stages, view initial/important/final states and play it to judge pacing. HTML viewing does not substitute for actual exported-image viewing.

Edit the same source draft when the picture reveals a problem, regenerate and re-view affected files or stages. Record substantive problems in the same issues.md. Finish when needed surfaces have been viewed and findings corrected and rechecked, or retained problems and viewing gaps have a stated reason and effect. File completion alone does not establish a high-quality result.

Only an explicit request for traceable review or audit adds formal review inputs and, for images, formal artifact evidence. Use author --formal-review to retain a stable source and use its returned v### for formal commands. Historical comparison/rollback requests use --checkpoint. These steps are independent of animation, complexity and output format; later ordinary edits continue the mutable current draft.

5. Deliver

Read finalization. Publish current JSON, paused offline HTML, report.json, report.md and requested formats:

node <skill-directory>/scripts/flowsvg.mjs finalize --workspace <task-workspace> --name <name>

If issues remain, summarize their effects in Agent-authored Markdown and pass --notes <summary.md>; the CLI copies it without interpreting or grading it. Verify final paths under .diagrams/<diagram-name>/ and scoped workspace cleanup. Keep history or work records only when requested.

Lead with the viewable result and links. Briefly explain how the structure answers the reader's question and any material assumptions or information choices. Report file completion, machine checks, actual viewing (files, dimensions and animation stages) and unperformed checks separately. Default reports state formal review is not requested; registered formal coverage describes evidence and viewing, with viewer issues reported separately from design acceptance.

Finish when files and notes are published and linked, viewing scope and remaining limitations are clear, and cleanup or requested retention is verified.

Recover interrupted work

On execution failure, retain the source draft, issues.md and available files. Read the relevant recovery branch, correct the cause within authorization, then retry the affected operation against current. A previous complete .recovery result retains its own source identity and cannot represent a changed draft. Preserve any prior published delivery.

After interruption, inspect remaining task files and continue the unfinished work; use explicit retained checkpoints only when needed. Stop repeated retries without new evidence and report the next action. On user cancellation, stop commands and retain work; resume when asked.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.