Przejdź do treści
Biblioteka skilli
Programowanie · Pisanie

readmex

/readmex

Tworzy, ocenia i upiększa README projektu: sprawdza twierdzenia i strukturę, dodaje obrazy główne oraz diagramy.

Licencja
MIT
Ostatnia zmiana repozytorium
12.11.2026

Ocena i bezpieczeństwo

100/ 100Klasa: dobry
Format25 / 25
Frontmatter z nazwą i opisem, poprawna nazwa, opis do 1024 znaków, nazwa zgodna z folderem.
Odkrywalność25 / 25
Opis mówi, kiedy użyć skilla, ma sensowną długość i jest w 3. osobie.
Treść35 / 35
Co najmniej 150 słów i do 500 linii, nagłówki, listy kroków, przykłady albo kod.
Higiena15 / 15
Bez zaślepek typu TODO, bez ścieżek Windows, z licencją.
Ryzykobrak sygnałów w skanie
Pliki w paczce
64
Skrypty
36
Komentarze HTML
0
Adresy URL w SKILL.md
0
Przestarzałość

Brak sygnałów. Repozytorium zmienione 12.11.2026, bez odwołań do wycofanych modeli i API.

Licencja

MIT, otwarta: można używać i zmieniać, z zachowaniem informacji o autorze.

Co sprawdzić przed instalacją
  • 36 skryptów w paczce. Przejrzyj je przed instalacją, bo Claude może je uruchomić.

Skrypty: bin/render-diagram.mjs, renderers/architecture/grid.mjs, renderers/architecture/labels.mjs, renderers/architecture/render-architecture.mjs, renderers/architecture/routing.mjs, renderers/shared/atomic-output.mjs, renderers/shared/brand-marks.mjs, renderers/shared/cli.mjs, renderers/shared/desktop-readability.mjs, renderers/shared/diagnostics.mjs, renderers/shared/engineering-profiles.mjs, renderers

Co robi i kiedy

Tworzy, ocenia i upiększa README projektu: sprawdza twierdzenia i strukturę, dodaje obrazy główne oraz diagramy.

Oryginalny opis (EN)

Create, review, or beautify a project README. Use for requests to write README.md, audit README claims or structure, score a README, or add README hero images and diagrams; supports quick creation, review, visual optimization, and the combined flow.

Jak działa

  1. Route
  2. Ask once, at the beginning
  3. Files and shared preparation
Sekcje z nagłówków SKILL.md, w oryginale · 59 linii, 739 słów
Podgląd SKILL.md
---
name: readmex
description: Create, review, or beautify a project README. Use for requests to write README.md, audit README claims or structure, score a README, or add README hero images and diagrams; supports quick creation, review, visual optimization, and the combined flow.
license: MIT
metadata:
  requirements: Python 3 and Node.js 18+; GIFs optionally use Pillow, ffmpeg, and rsvg-convert or macOS sips.
---

# readmex

Help project authors write a clear, concise, attractive README. Select only the requested pipeline; load its instructions when needed. The combined flow is **quick creation → review → optimization**. Do not produce the old interactive project overview or insert its blocks into the README.

## Route

| User request | Pipeline |
| --- | --- |
| “Write a README” / “写个 README”, with no scope | Ask “要快速简洁的,还是完整清晰并且有配图的?” in the single opening question batch; quick selects quick creation, complete selects the combined flow |
| Quick / “快速写一份” | [quick-create.md](pipelines/quick-create.md) |
| Review, check, score, “为什么没人用” **about a README** | [review.md](pipelines/review.md) |
| Beautify, hero image, architecture or workflow image | [optimize.md](pipelines/optimize.md) |
| Complete, full process, “一条龙” | All three, in order |

Use the target repository given by the user, otherwise the workspace root. If quick creation encounters `README.md`, route to review instead of replacing it. An explicit request to **rewrite** authorizes replacement. In the combined flow an existing README similarly skips creation and enters review → optimization. Do not back up a README before changing it.

## Ask once, at the beginning

For a known review or optimization route, scan and inspect relevant code before asking. Quick creation reads the project directly and **does not run the scanner**. For an ambiguous “write a README”, inspect necessary files first, then include route selection and questions conditional on choosing the complete flow in **one message**; scan only after the selected route needs it.

Infer repository, project type, existing brand, and publication targets from files. Combine only applicable questions into that one message:

- Route selection only for an ambiguous “write a README”.
- Review facts, at most six questions: always ask whether adoption is personal or company procurement, which alternatives readers already know, and what legal or policy content must appear. Ask who the readers are and the project's actual status only when repository evidence is inconclusive.
- Standalone optimization: ask whether to refresh the whole page or produce assets only. The combined flow defaults to a whole-page refresh.
- When planning a hero: “是否可以调用生图工具生成素材?”
- When planning a hero or workflow image: “是否可以把头图和流程图做成 GIF 动图?”

Reader identity and adoption are separate facts: “individual developer” alone does not establish whether adoption is personal or requires company procurement. Use an explicit adoption answer rather than inferring it from the reader's job.

Image generation and GIFs each require an explicit affirmative answer; no answer or refusal means static SVG only. Existing explicit choices count as answers. Do not ask further design or factual questions during execution. During review or optimization, downgrade or delete newly discovered unsupported claims and record them in the scorecard. The one later approval gate is **standalone review**: deliver the diagnosis and wait for agreement before rewriting `README.md`. Combined execution already authorizes the rewrite. Linux GIF installation may require the user to enter a sudo password in their own terminal; this is an installation interaction, never a request to disclose the password.

## Files and shared preparation

Resolve helper paths from the installed skill directory, not the target's working directory. Before producing intermediates, append the exact line `.readmex/` to the target `.gitignore` if absent; create `.gitignore` when needed and preserve its other contents.

For review or optimization, run once per invocation and reuse the facts across pipelines:

```bash
python3 <skill>/scripts/scan.py <target> --output <target>/.readmex/scan.json
```

Read entrypoints and code as well; the scan is an inventory, not evidence of behavior. The scanner also prints JSON to stdout. Language follows the user's instruction, then an existing README, then English.

| Location in the target | Contract |
| --- | --- |
| `README.md` | Final project page at the repository root |
| `readme/assets/` | Only files referenced by the README; commit these assets |
| `.readmex/scan.json` | Shared repository inventory |
| `.readmex/review/audit.txt`, `scorecard.md` | Mechanical findings and evidence-based diagnosis |
| `.readmex/optimize/story.md` | Five-line project story and frozen visual specification |
| `.readmex/optimize/diagrams/architecture.json`, `workflow.json` | Authored renderer input |
| `.readmex/optimize/imagegen/` | `hero-layout.svg`, `hero-subject.png`, `hero-prompt.txt` |
| `.readmex/optimize/motion/` | Motion specification JSON; delete rendered frames after production |
| `.readmex/optimize/preview/` | Previews at 900 and 360 pixels wide |

Published assets may include `hero.svg` or a composited `hero.png`, `hero.gif` alongside `hero.svg`, `architecture.svg`, `workflow.svg` and optional `workflow.gif`, and referenced `section-*.svg`. Keep unpublished sources and previews in `.readmex/`. Clean up only assets owned by this run that are no longer referenced; preserve unrelated files.
Pliki w paczce (64)
  • LICENSE 1 KB
  • SKILL.md 6 KB
  • THIRD_PARTY_NOTICES 5 KB
  • bin/render-diagram.mjs 2 KB
  • examples/agent-tool-call.workflow.json 1 KB
  • examples/web-app.architecture.json 1 KB
  • pipelines/optimize.md 6 KB
  • pipelines/quick-create.md 2 KB
  • pipelines/review.md 3 KB
  • references/architecture-layout-repair.md 2 KB
  • references/authoring-contract.md 4 KB
  • references/authoring-defaults.md 3 KB
  • references/brand-marks.md 1 KB
  • references/hybrid-svg-production.md 2 KB
  • references/motion-production.md 3 KB
  • references/optimize-visual.md 3 KB
  • references/readme-skeletons.md 3 KB
  • references/renderer-authoring.md 2 KB
  • references/repository-authoring.md 2 KB
  • references/review-evidence.md 4 KB
  • references/review-scorecard.md 6 KB
  • renderers/architecture/grid.mjs 2 KB
  • renderers/architecture/labels.mjs 10 KB
  • renderers/architecture/render-architecture.mjs 47 KB
  • renderers/architecture/routing.mjs 54 KB
  • renderers/shared/atomic-output.mjs 69 KB
  • renderers/shared/brand-marks.mjs 27 KB
  • renderers/shared/cli.mjs 20 KB
  • renderers/shared/desktop-readability.mjs 7 KB
  • renderers/shared/diagnostics.mjs 7 KB
  • renderers/shared/engineering-profiles.mjs 7 KB
  • renderers/shared/generated-brand-marks.mjs 160 KB
  • renderers/shared/generated-validators.mjs 240 KB
  • renderers/shared/geometry.mjs 82 KB
  • renderers/shared/i18n.mjs 42 KB
  • renderers/shared/layout-report.mjs 1 KB
  • renderers/shared/legend.mjs 8 KB
  • renderers/shared/output-path.mjs 19 KB
  • renderers/shared/path-semantics.mjs 32 KB
  • renderers/shared/portable-path.mjs 12 KB
  • renderers/shared/repository-evidence.mjs 20 KB
  • renderers/shared/repository-location.mjs 3 KB
  • renderers/shared/route-quality.mjs 23 KB
  • renderers/shared/sidecar-path.mjs 1 KB
  • renderers/shared/spatial-grid.mjs 2 KB
  • renderers/shared/static-svg.mjs 3 KB
  • renderers/shared/svg-theme.css 8 KB
  • renderers/shared/text-fit.mjs 5 KB
  • renderers/shared/utils.mjs 8 KB
  • renderers/shared/validator.mjs 3 KB
  • renderers/workflow/README.md 1 KB
  • renderers/workflow/render-workflow.mjs 1 KB
  • renderers/workflow/workflow-compiler.mjs 191 KB
  • renderers/workflow/workflow-migration-geometry.mjs 5 KB
  • schemas/architecture.schema.json 6 KB
  • schemas/common.schema.json 8 KB
  • schemas/workflow.schema.json 10 KB
  • scripts/readme_factor_audit.py 16 KB
  • scripts/readme_view_audit.py 9 KB
  • scripts/render_motion_gif.py 26 KB
  • scripts/scan.py 8 KB
  • scripts/data/dependency_files.json 1 KB
  • scripts/data/ignore_patterns.json 1 KB
  • scripts/data/language_mapping.json 2 KB

Wersje

W bazie nie ma innych skilli o nazwie readmex.

Migawka bazy z 13.11.2026 · wersja paczki 24e4fd8192a8 (skrót SHA-256 całej paczki)

Pobrania w ostatnich 30 dniach: 0 · ranking

Zgłoś problem z tym skillem: szkodliwa treść, błędny opis albo prośba autora o usunięcie.

Podobne skille

  • 95readme
    Tworzy lub aktualizuje plik README.md projektu na podstawie jego kodu i dokumentacji.
  • 96create-readme
    Generuje lub aktualizuje plik README.md albo CHANGELOG.md projektu.
  • 100readme
    Tworzy lub aktualizuje README projektu: co to jest, jak zainstalować, używać i zbudować, sprawdzone z repozytorium.
  • 100readme-orchestrator
    Koordynuje projektowanie README: wybiera tryb pracy, rozstrzyga konflikty stylu między pięcioma skillami i weryfikuje wynik.
  • 90create-readme
    Tworzy plik README.md dla nowego repozytorium lub przy jego braku, analizując dokumentację i strukturę katalogów projektu.
  • 100write-readme
    Pisze lub audytuje README projektu, dobiera sekcje do typu projektu i sprawdza linki, nagłówki oraz fragmenty kodu.

Uwagi i zgłoszenia

Wiadomość trafi do Bartosza Sokolińskiego. Adres e-mail jest potrzebny tylko do odpowiedzi. Szczegóły w polityce prywatności.