@stashwiseapp/mcp

Search your Stashwise library + wiki from any AI agent — Claude Code, Codex CLI, Cursor, Claude desktop.

Install

Drop one block into your agent host's MCP config — no separate install step.

Claude Code (~/.claude/settings.json):

{
  "mcpServers": {
    "stashwise": {
      "command": "npx",
      "args": ["-y", "@stashwiseapp/mcp"]
    }
  }
}

Cursor (~/.cursor/mcp.json) and Claude Desktop (~/Library/Application Support/Claude/claude_desktop_config.json) use the same JSON shape.

Codex CLI (~/.codex/config.toml):

[mcp_servers.stashwise]
command = "npx"
args = ["-y", "@stashwiseapp/mcp"]

One-time login

npx -y @stashwiseapp/mcp auth

Opens stashwise.co/cli in your browser. Sign in, click Authorize, and the CLI stores a long-lived token in your OS keychain. The agent picks it up automatically on its next tool call.

What the agent sees

One tool, search_stashwise(query, k, scope). Returns ranked snippets from your saved content with citations:

{
  "results": [
    {
      "kind": "content",
      "id": "...",
      "title": "Why HNSW is the default vector index",
      "snippet": "HNSW (Hierarchical Navigable Small World) graphs trade...",
      "source_url": "https://...",
      "source_platform": "article",
      "score": 0.87,
      "citation": "Why HNSW is the default vector index — example.com",
      "saved_at": "2026-05-12T18:22:13+00:00"
    }
  ],
  "query": "vector index choice",
  "retrieval_ms": 41
}

Diagnostics

npx -y @stashwiseapp/mcp doctor

Prints config, token validity, and backend reachability. Use when the agent reports it can't reach your Stashwise content.

Manage tokens

Visit stashwise.co/account/mcp to see all authorized agents and revoke any of them.

Environment

Variable	Default	Purpose
STASHWISE_API_URL	https://stashwise-api.fly.dev/api/v1	Backend base URL
STASHWISE_WEB_URL	https://stashwise.co	Webapp base URL (used in auth flow)

License

MIT