
AppScreenshots
io.github.horace68Published by GitHub account horace68v0.1.0Updated Nov 4, 2026
Create and edit app store screenshot projects with AI, then export in the web editor.
Overview
Remote MCP server that lets an AI agent create and edit app store screenshot projects in AppScreenshots, then hand off an editor link for export.
- What you can do
- Create and edit app store screenshot projects from authorized assets, choosing available templates and keeping headlines short.
- Work with projects and individual assets the user has authorized, optionally allowing new projects.
- Review designs with the client's own browser tools, inspecting screenshots and correcting clipping, cropping or layout issues.
- Return the project editor link so the user can export on the website.
- When to use it
- Useful when you want an assistant to draft and refine App Store or Google Play screenshot sets from screenshots already uploaded to AppScreenshots, and you are willing to finish visual checks and export in the web editor.
- Requirements
- Remote Streamable HTTP endpoint at appscreenshots.net with OAuth sign-in to an AppScreenshots account; no API keys or database credentials. The optional Skill installer needs Git and Node.js 20+. Browser and image tools come from the client, not the server. Project, template and export entitlements apply.
Installation
In SourceWeft
- Open AppScreenshots in the dashboard and add it to a workspace.
- Connect the required credentials. They are stored privately and never shown on public pages.
- Enable the server for the chats that should use its tools.
Web executable via Streamable HTTP. Remote servers run from the web runtime once configured in a workspace.
Use with an agent
Paste this prompt into Claude Code, Codex, Cursor or another agent — or into SourceWeft chat. The agent reads this server's install guide, shows you its publisher, endpoint and sign-in, and adds it to your client once you agree. Keys never go into a command.
Read https://sourceweft.com/mcp/io.github.horace68%2Fappscreenshots/install.md and connect the MCP server it describes to my client. Before changing any config, show me its publisher, endpoint and sign-in method and wait for my OK. Never put an API key in a command or in a file I would commit; ask me where it should come from.Other MCP clients
Add this to your client's mcpServers config. The client completes OAuth sign-in on first connect.
{
"mcpServers": {
"appscreenshots": {
"type": "http",
"url": "https://appscreenshots.net/api/mcp"
}
}
}README
AppScreenshots MCP
简体中文 · Client setup · Tools · Troubleshooting
Let your AI agent create editable app screenshot projects, work with your authorized assets, review and correct designs using available client browser tools, then hand off the editor link for website export in AppScreenshots.
Preview service is live. On September 18, 2026, a protocol client completed OAuth, created and edited a two-frame project, opened it in the website editor, and exported two PNGs. Individual Claude Code, Codex, Gemini CLI, Kimi Code, ZCode and ChatGPT integrations have not yet been certified. Until the website's loopback callback fix is deployed, clients using a
127.0.0.1redirect URI may fail authorization; see troubleshooting.
This public repository contains connection examples, a small installation helper, and an optional agent Skill. The MCP server runs within AppScreenshots. Website, authentication, database, billing and rendering backend code remain in their existing private deployment. The helper is not a local stdio server.
Connect the MCP service
Remote Streamable HTTP endpoint:
Choose your client and follow its OAuth flow. Sign in with your AppScreenshots account, choose projects/assets, and optionally allow new projects. No database credentials or service API keys are required.
Claude Code
Open /mcp in Claude Code and authorize AppScreenshots.
Codex
For Gemini CLI, Kimi Code, ZCode and ChatGPT, see client setup. MCP is sufficient to call the tools; the Skill below provides a reusable design and preview workflow.
Install the optional Skill
Requires Git and Node.js 20+. No npm dependencies or install hooks are used.
For Claude Code, replace the last line with:
The installer copies skills/appscreenshots into ~/.agents/skills/appscreenshots for Codex or ~/.claude/skills/appscreenshots for Claude Code. It refuses to overwrite an existing directory. Restart the agent to discover the Skill, then use $appscreenshots in Codex or /appscreenshots in Claude Code. Connect and authorize MCP separately.
For project-local or other compatible skill hosts, choose the parent skill directory explicitly:
Use the directory required by your client. The installer does not auto-configure other clients or modify existing MCP settings. To update a customized Skill, compare it with this repository and merge your edits; for a clean update, move the old directory aside and rerun installation. To uninstall, remove only the installed appscreenshots skill directory. Revoke server access separately in website settings.
Connection helper
config prints a snippet to merge into existing settings. doctor checks public OAuth discovery, sends no credentials, and exits nonzero if discovery is unavailable. It does not authenticate or test rendering. The npm package name is not published to the npm registry; use this Git repository.
Try a task
Use my authorized screenshots to create an English App Store screenshot project. Pick an available template, keep the headlines short, open the editor with available browser tools, inspect screenshots, correct clipping/cropping/layout issues, then give me the project link so I can export on the website.
Upload screenshots on the website first and authorize the project or individual assets. Chat attachments are not automatically available to MCP. Tool reference explains editing, retries, previews and downloads.
Availability and data
The default flow does not require cloud rendering: MCP saves projects, the agent inspects the website using its own browser tools, and the user exports in the editor. Browser access is supplied by the client, not installed by this Skill. Website login is separate from MCP OAuth. If browser/image tools are unavailable, visual verification remains pending. Project, template and export entitlements still apply. See the browser review guide.
Manage or revoke access in Settings → Connected AI tools. This repository does not implement purchases, deletion, hosted AI generation or app-store publishing. Security and scope.
Development
Tests exercise configuration generation, installation isolation, overwrite protection, CLI failures and discovery diagnostics. CI runs on Linux, macOS and Windows with Node.js 20 and 22. A protocol client has completed the hosted create, editor preview and export flow; individual client integrations and asset upload remain unverified. See the compatibility matrix.
MIT license covers this repository's helper, examples and Skill; it does not grant access to the hosted service or rights to user assets, paid templates or the website backend.
Source: README.md at commit d8118fd
Tools
0Version history
1- v0.1.0LatestOct 14, 2026
Score
85% of the applicable checks met
How complete this listing is, computed from the registry entry, the endpoint and public sources. Not a rating of the server and not a statement that it is safe or working today: reachability is a dated check, and no installs or reviews go into the score.
- Reached in the last 30 daysRequired25 pts
- Auth declaredRequired10 pts
- Tools listed15 pts
Answer tools/list without a credential, or ship a package that starts in a sandbox without one.
- DescriptionRequired15 pts
- Logo10 pts
- Documentation15 pts
- License10 pts
