A
akkijp/app-spec-crawler-skill
Claude Code skill: point it at any web app you're logged into and get a screen spec with screenshots
View on GitHub ↗Link checked 2026-11-12 · HTTP 200 · opens the vendor's own site
Install & config
Client config (built from the install command this repo gives)
{
"mcpServers": {
"akkijp-app-spec-crawler-skill": {
"command": "npx",
"args": ["skills"]
}
}
}Parsed from the README we fetched on 2026-11-11; we have not executed it.
command/args come from the install command the repository states. When it is ambiguous, the project README wins.
VS Code (.vscode/mcp.json - the key is servers)
{
"servers": {
"akkijp-app-spec-crawler-skill": {
"command": "npx",
"args": ["skills"]
}
}
}Same server, but VS Code keeps it under <code>servers</code>.
Claude Code (command line)
claude mcp add --transport stdio akkijp-app-spec-crawler-skill -- npx skills
Claude Code does not read a config file; add the server with this command.
Key names differ by client: Clients & CLI
- Star count
- 0 ★ · akkijp/app-spec-crawler-skill · read from the GitHub search API · (checked 2026-11-13 20:00:50)
- Source
- read directly from its GitHub page · type: open-source project
- Harvest history
- first seen 2026-11-12 · last seen 2026-11-14 · seen 3× · listed by 2 sources
- Score
- score 203 · 2 sources · repo
- Install
npx skills
FAQ (answers come from our records)
How do I install it?
The repository provides this install command:
npx skills. It comes from the repository itself; we cannot guarantee it runs as-is in your client.Who is it for?
We filed it under these tags: Agent 技能, 浏览器自动化. The tags come from the entry itself, not a taxonomy we imposed.
Who maintains it?
repository
akkijp/app-spec-crawler-skill；first catalogued 2026-11-12；last seen 2026-11-14。There is no official certification or quality rating here — we record sources and dates only.Point Claude Code at any web app you're logged into. Get the spec — every screen
以下摘录自该仓库的 README（片段） · 在 GitHub 读完整 README →
English · 日本語
https://github.com/user-attachments/assets/0f28552f-e581-4f41-b838-a610962fed5f
Recorded on an earlier run of the WordPress sample (51 screens, 381 spec items). With sound.
I inherited an admin panel with no documentation. This Claude Code skill walks every screen in your own logged-in Chrome, takes a full-page screenshot of each, and has Claude write the documents you wish had come with the system.
## Use it for
Internal tools nobody documented
Taking stock of a SaaS before you migrate off it
Onboarding: "what does this system even do?"
## What you get
System guide — one page for newcomers: what the system does, a map of the screens, the main workflows, the rules
Screen spec — every screen, with numbered spec items. Each item says whether it was Verified on screen, Inferred, or still To confirm, and links to the screenshot it came from
Both are plain HTML. Open them in a browser, zip them, hand them over.
See a real one → A WordPress admin: 50 screens, 376 spec items, nothing written by hand.
The documents are written in the language you talk to Claude in. Five languages are supported: English, Japanese, Simplified Chinese, Spanish and Korean. In any other language the frame is in English. The Chinese, Spanish and Korean wording is machine-translated; corrections are welcome.
## Read-only
The crawler opens pages by URL and looks. It has no code that types into a field or submits a form.
It never clicks Save, Delete, Approve, Send or anything like them. The only clicks are on in-page tabs.
It does not open URLs that contain words such as logout, delete, approve or export.
It does not open links that carry an anti-forgery token (_wpnonce, csrf_token, authenticity_token…) or an action parameter that is not known to be read-only. Some apps change data on a plain link — WordPress "Trash" and "Activate" are examples — and this is how such links are recognised.
Every URL it decided not to open is listed, with the reason, in safety_report.json.
What it cannot know: an app can change something merely because a page was opened. WordPress, for example, creates a draft the moment "Add Post" loads. If you know of such a screen, exclude it with --exclude. On a system you cannot afford to disturb, ask the people who built it first.
Details: how the safety rules work, the full policy.
## Install
bash
npx skills add akkijp/app-spec-crawler-skill
That copies the skill into your Claude Code skills folder. Then ask Claude Code to run the app-spec-crawler doctor, and it tells you what else is needed.
Or paste this into Claude Code. Claude installs the skill with git, puts the command on your PATH, and checks that it can run.
text
Install the Claude Code skill app-spec-crawler.
1. git clone https://github.com/akkijp/app-spec-crawler-skill.git ~/.claude/skills/app-spec-crawler
(if it already exists, do not overwrite it — ask me)
2. mkdir -p ~/.local/bin && ln -sf ~/.claude/skills/app-spec-crawler/bin/app-spec-crawler ~/.local/bin/app-spec-crawler
3. Run app-spec-crawler doctor and show me the result as it is.
https://github.com/user-attachments/assets/0f28552f-e581-4f41-b838-a610962fed5f
Recorded on an earlier run of the WordPress sample (51 screens, 381 spec items). With sound.
I inherited an admin panel with no documentation. This Claude Code skill walks every screen in your own logged-in Chrome, takes a full-page screenshot of each, and has Claude write the documents you wish had come with the system.
## Use it for
Internal tools nobody documented
Taking stock of a SaaS before you migrate off it
Onboarding: "what does this system even do?"
## What you get
System guide — one page for newcomers: what the system does, a map of the screens, the main workflows, the rules
Screen spec — every screen, with numbered spec items. Each item says whether it was Verified on screen, Inferred, or still To confirm, and links to the screenshot it came from
Both are plain HTML. Open them in a browser, zip them, hand them over.
See a real one → A WordPress admin: 50 screens, 376 spec items, nothing written by hand.
The documents are written in the language you talk to Claude in. Five languages are supported: English, Japanese, Simplified Chinese, Spanish and Korean. In any other language the frame is in English. The Chinese, Spanish and Korean wording is machine-translated; corrections are welcome.
## Read-only
The crawler opens pages by URL and looks. It has no code that types into a field or submits a form.
It never clicks Save, Delete, Approve, Send or anything like them. The only clicks are on in-page tabs.
It does not open URLs that contain words such as logout, delete, approve or export.
It does not open links that carry an anti-forgery token (_wpnonce, csrf_token, authenticity_token…) or an action parameter that is not known to be read-only. Some apps change data on a plain link — WordPress "Trash" and "Activate" are examples — and this is how such links are recognised.
Every URL it decided not to open is listed, with the reason, in safety_report.json.
What it cannot know: an app can change something merely because a page was opened. WordPress, for example, creates a draft the moment "Add Post" loads. If you know of such a screen, exclude it with --exclude. On a system you cannot afford to disturb, ask the people who built it first.
Details: how the safety rules work, the full policy.
## Install
bash
npx skills add akkijp/app-spec-crawler-skill
That copies the skill into your Claude Code skills folder. Then ask Claude Code to run the app-spec-crawler doctor, and it tells you what else is needed.
Or paste this into Claude Code. Claude installs the skill with git, puts the command on your PATH, and checks that it can run.
text
Install the Claude Code skill app-spec-crawler.
1. git clone https://github.com/akkijp/app-spec-crawler-skill.git ~/.claude/skills/app-spec-crawler
(if it already exists, do not overwrite it — ask me)
2. mkdir -p ~/.local/bin && ln -sf ~/.claude/skills/app-spec-crawler/bin/app-spec-crawler ~/.local/bin/app-spec-crawler
3. Run app-spec-crawler doctor and show me the result as it is.
Listed badge (optional)
Copy this line into the README if you want the listing mark. Optional, and it does not affect being listed.
[](https://anclum.com/en/s/akkijp-app-spec-crawler-skill/)
- Source
- github · https://github.com/akkijp/app-spec-crawler-skill
- Browse by field
- Agent skills, Browser automation
- Listed by
- 2 independent sources
- added
- today
More like this
Filed under
Every tag opens the full list for that filter. Counts are the live catalogue figures.