Wedges
A remote MCP server that guides agents through taste extraction exercises to produce a portable taste profile (style guide, worldview, voice patterns, etc.), enabling agents to serve your work without flattening it.
README
Wedges
The agent edition of Both Hands Full.
Both Hands Full is a set of exercises that help humans protect their taste, voice, and judgment in the age of synthetic everything. Wedges is the inversion: point your agent at the same exercises and walk away with a portable taste profile — a style guide, worldview, glossary, voice patterns, and irreducible list — that any agent can load to serve your work without flattening it.
It's a public, no-auth remote MCP server. Connect Claude Code, Codex, ChatGPT, Cursor — same URL.
Hold critique in one hand. Hold curiosity in the other. Keep walking.
Connect
Claude Code:
claude mcp add --transport http wedges https://wedges.dev/api/mcp
Any client (.mcp.json / config):
{
"mcpServers": {
"wedges": { "type": "http", "url": "https://wedges.dev/api/mcp" }
}
}
Then, in a session: "Run the Wedges taste extraction." The agent self-drives via the start_wedges prompt.
What it exposes
| Tool | What it does | LLM? |
|---|---|---|
list_exercises |
The exercise catalog | no |
mirror_booth |
Voice-drift analysis of a paragraph | yes |
taste_audit |
Visual taste analysis of up to 12 image URLs | yes (vision) |
selector_pressure_test |
Deterministic taste profile from 10 rounds of choices | no |
name_irreducibles |
Structure the "things AI can't eat" list | no |
export_profile |
Assemble the portable taste profile (markdown + JSON) | no |
Resources: wedges://catalog, wedges://profile. Prompt: start_wedges.
Keys & privacy
- The LLM tools use the server's
ANTHROPIC_API_KEYby default. To use your own, passanthropic_api_keyas a tool argument. - Nothing is stored. Profiles are handed back to your agent and forgotten. No accounts, no database, no auth.
- The two LLM tools (
mirror_booth,taste_audit) are rate-limited — ~10/min per IP on the shared server key, ~60/min when you bring your own. The deterministic tools are unthrottled. (The limiter is in-memory/per-instance — a v1 guardrail; seelib/rate-limit.ts.)
Develop
npm install
npm run dev # http://localhost:3000/api/mcp
npm run verify # typecheck + the MCP smoke test (needs the dev server running)
The exercise logic is ported from the Both Hands Full app; see lib/. The MCP server is one route: app/api/[transport]/route.ts.
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.