Prodpoke
@prodpoke
Prodpoke について
概要はまだありません
基本情報
設定
以下の設定を使って、このサーバーを MCP 対応クライアントに追加してください。
{
"mcpServers": {
"prodpoke": {
"command": "npx",
"args": [
"mcp-remote",
"https://prodpoke.com/v1/mcp",
"--header",
"Authorization: Bearer ${PRODPOKE_API_KEY}"
],
"env": {
"PRODPOKE_API_KEY": "pp_your_key_here"
}
}
}
}ツール
2Talk to ProdPoke like a QA tester — scan a URL, test a flow, investigate findings, set session preferences. Stateful via `session_id`.
Check your remaining scan credits.
概要
What is Prodpoke?
ProdPoke is a remote, stateful MCP server that runs real-browser QA for any website. It launches headless Playwright sessions on ProdPoke infrastructure, finds bugs (accessibility, SEO, performance, visual regressions, broken CTAs, and more), and remembers conversation context across turns—all without installing anything locally.
How to use Prodpoke?
Get a free API key (prefixed pp_) from the ProdPoke dashboard. Paste the streamable-http configuration into your MCP client (Claude Desktop, Cursor, Claude Code) with the key in the Authorization header. Restart the client and chat with the chat tool—for example, "Scan https://mysite.com and tell me what's broken".
Key features of Prodpoke
- Remote Playwright browser on hosted infrastructure.
- No local Chromium or container management needed.
- Stateful multi-step sessions via
session_id. - Structured findings (accessibility, SEO, performance, flows).
- Streamable HTTP transport (with stdio fallback via bridge).
- Session context persists across client restarts.
Use cases of Prodpoke
- Scan any public URL for accessibility, SEO, and performance issues.
- Test user flows (signup, checkout) and detect broken CTAs.
- Compare competitor SEO headers or landing pages side-by-side.
- Investigate findings iteratively by asking follow-up questions.
- Check credit balance and track scan costs.
FAQ from Prodpoke
What infrastructure does ProdPoke require?
Nothing is installed on your machine. Scans run on ProdPoke servers—just a standard MCP client and an API key are needed.
How are sessions and data handled?
The chat tool is stateful: each response returns a session_id. Passing it back continues the conversation with full context. Sessions persist for the lifetime of the API key.
What transport and authentication are used?
Streamable HTTP with a Bearer token (Authorization: Bearer pp_...). Clients that only support stdio can use the mcp-remote bridge as a proxy.
Are there any request timeouts or limits?
The MCP call times out at 120 seconds per step. Long workflows should be split into multiple chat turns using the session_id. Standard scans cost ~10 credits; deeper flows 400–500 credits.
How does pricing work?
Free tier grants 500 credits. Paid plans: Watch ($15/mo, 10,000 credits), Watch Pro ($49/mo, 50,000 credits). Top-up packs from $3. Insufficient credits return an explicit error message.
「その他」の他のコンテンツ
Awesome Mlops
visengerA curated list of references for MLOps
FastMCP v2 🚀
jlowin🚀 The fast, Pythonic way to build MCP servers and clients.
MCP Go 🚀
mark3labsA Go implementation of the Model Context Protocol (MCP), enabling seamless integration between LLM applications and external data sources and tools.
Core Philosophy: Connect, Unify, Respond
mindsdbDelegate anything. It comes back done.
Codelf
unbugA search tool helps dev to solve the naming things problem.
コメント