Claude Code in 2 minutes
Project-scoped `.mcp.json`, expected output, and no-lock-in Cursor link.
Pick Claude Code, Cursor, or Codex, enter the memory library id,
and copy a config that runs npx answer-engine-mcp
against https://engine.answeragent.ai. The source URL field frames what
you will ingest after the server connects; it does not call an API
from this public page.
The generated JSON mirrors the Claude Code and Cursor install guides:
one server named answer-engine, one npx command, one
scoped library, and the corrected Answer Engine host.
Project-scoped Claude Code config. Use user scope only when you want the same memory in every repo.
{
"mcpServers": {
"answer-engine": {
"command": "npx",
"args": [
"answer-engine-mcp"
],
"env": {
"ANSWER_ENGINE_API_KEY": "ae_live_your_key_here",
"ANSWER_ENGINE_API_URL": "https://engine.answeragent.ai",
"ANSWER_ENGINE_LIBRARY": "your-library-id"
}
}
}
} npx answer-engine-mcp
Ask your agent to use the answer-engine MCP server after it connects.
Use the guides when you want the exact client-specific walkthrough, expected output, and troubleshooting checks.
Project-scoped `.mcp.json`, expected output, and no-lock-in Cursor link.
Workspace `.cursor/mcp.json`, expected output, and no-lock-in Claude Code link.
The evaluation rubric for scoped, source-aware, inspectable memory.
It creates a valid Answer Engine MCP config JSON block and the matching npx answer-engine-mcp install command for Claude Code, Cursor, or Codex-style MCP clients.
No. It uses an ae_live_ placeholder so the generated config is safe to copy before you create or paste a real scoped key.
The generated config always uses https://engine.answeragent.ai for the API base URL and references https://engine.answeragent.ai/mcp for hosted HTTP MCP.