wellworn
OfficialEvaluates Clerk as an authentication provider, noting when it is the best choice compared to alternatives and surfacing relevant traps.
Surfaces known breakages and fixes for Drizzle ORM, including adapter-specific traps such as requiring explicit schema configuration.
Provides stack recommendations, alternatives, and known traps for Next.js applications, such as choosing an auth solution for a solo developer.
This server does not currently have a deployable release on Glama. Follow the repository's setup instructions to install and configure it for your MCP client, including any required credentials and local dependencies.
Wellworn tells your agent which stack, library, skill, or design system to use, what breaks at the version in your lockfile, and who verified it, in under 800 tokens. It works without an account or key.
Without Wellworn
Your agent picks from training data. It reaches for the library it saw most in 2024, a version that no longer exists, or a package that was deprecated last spring. Sonatype measured 27.76% of AI dependency upgrade suggestions pointing to non-existent, deprecated, or unsafe versions.
Related MCP server: Vibe Coder MCP Server
With Wellworn
Before it adds a dependency or lays out a screen, the agent calls recommend, compare, alternatives, traps, design, or docs and gets a verdict:
VERDICT auth for a Next.js app, solo dev, free tier
PICK Better Auth 1.7: own your users table; org and api-key plugins; no per-MAU bill
ALT Clerk: hosted UI and SSO in a week; wins when you accept per-MAU pricing
AVOID NextAuth: v5 migration stalled since 2025; session callback traps
TRAPS (2)
[a1f4c0d2] major: role changes wait for the 5 min cookie cache
[c9e3b7a1] minor: Drizzle adapter needs the schema passed explicitly
SKILL npx skills add WellWorn-dev/wellworn/skills/better-auth-nextjs
VERIFIED 2026-09-05 by youcef · recheck 2026-12-04 · sources 3
remaining: 59Every verdict names a reviewer, a date, the version tested, and an expiry. A nightly job re-reads the registries; when a major version moves, the verdict is served with a [RECHECK] flag instead of going stale.
Works with
Any client that speaks Streamable HTTP takes the same URL: https://mcp.wellworn.dev/mcp. No key: 60 calls a day per IP. Free key: 2,000 a month per organization, sent as Authorization: Bearer ww_.... Clients that only speak stdio bridge with npx mcp-remote https://mcp.wellworn.dev/mcp.
Client | Setup | Guide | |
Claude Code |
| ||
Cursor |
| ||
VS Code (Copilot) |
| ||
Windsurf |
| ||
Codex CLI |
| ||
Gemini CLI |
| ||
Cline |
| ||
Zed |
| ||
opencode |
|
Claude Code plugin
The plugin adds the server, a skill that tells Claude when to ask it, and a hook that prints known traps whenever a dependency manifest is about to change. The hook is advisory; it never blocks an edit and never sends file contents.
claude plugin marketplace add WellWorn-dev/wellworn
claude plugin install wellwornCursor
{ "mcpServers": { "wellworn": { "url": "https://mcp.wellworn.dev/mcp" } } }Gemini CLI
{ "mcpServers": { "wellworn": { "httpUrl": "https://mcp.wellworn.dev/mcp" } } }Tools
Tool | Returns |
| pick, alternative and when it wins, avoid, traps, skill, provenance |
| the same block for 2 to 4 named options |
| open-source or self-hosted alternatives to a SaaS or library |
| known breakages for a library at a version; pass |
| the SKILL.md procedure or its install command |
| design system, tokens, layout rules, anti-patterns for a screen |
| official docs URL, llms.txt, latest version |
| report a breakage and its fix to a reviewer (needs a free key) |
Plain text without MCP: curl "https://mcp.wellworn.dev/api/traps?library=better-auth&version=1.7.3".
What is in this repository
Path | What | License |
| the MCP server (TypeScript, | MIT |
| schema and queries, the compact renderer with its size contract, the Redis limiter | MIT |
| the Claude Code plugin (MCP config, skill, PreToolUse hook) | MIT |
| installable skills in the Agent Skills format: | MIT |
| the verdict format, JSON Schema | CC0 |
| the public corpus, one YAML per verdict, mirrored from the review database | CC BY-SA 4.0 |
| manifest for the MCP registry |
The web platform, the review tooling, and the signal pipeline that keeps verdicts fresh run at https://wellworn.dev and are not part of this repository.
Running the server yourself
pnpm install
cp packages/mcp/.env.example packages/mcp/.env # DATABASE_URL, REDIS_URL
DATABASE_URL=... pnpm --filter @wellworn/db migrate
pnpm --filter @wellworn/mcp devTests: pnpm test (Postgres and Redis required).
Contributing
Contributions are welcome: verdicts, traps, skills, client setup pages, and server fixes. Start with an issue template (verdict, trap, bug) or a pull request; see CONTRIBUTING.md and CODE_OF_CONDUCT.md. Traps are fastest through submit_trap; verdicts and corrections arrive as pull requests against verdicts/ and go through a human reviewer.
License
MIT for the code. Corpus CC BY-SA 4.0. Spec CC0.
This server cannot be deployed
Maintenance
Related MCP Connectors
Hosted MCP server and Agent Skills for specs, engineering threads, and project memory.
Search and inspect source-pinned agent skills, plugins and tools through a free public MCP.
Verified component catalog over remote MCP: agents pick component ids and get a stamped project repo that builds. Connect at https://www.botkelp.com/mcp. Auth: MCP OAuth 2.1, or a Bearer bk_live_ API key; initialize and tools/list are open.
Hosted MCP server for live public-data APIs and Skills for AI agents.
Related MCP Servers
- AlicenseNot gradedqualityAmaintenanceEnables AI assistants and developers to research, plan, generate requirements and starter projects, map codebases, manage tasks, curate context, and execute multi-step software development workflows through MCP-compatible clients.18 npmMIT
- AlicenseNot gradedqualityAmaintenanceEnables AI assistants to research, plan, generate requirements, create starter projects, map codebases, curate context, manage tasks, and run multi-step development workflows through MCP-compatible clients.18 npmMIT
- AlicenseNot gradedqualityAmaintenanceEnables AI assistants to perform software development tasks such as codebase analysis, research, planning, requirements and user-story generation, task management, project scaffolding, and multi-step workflow execution. It integrates with MCP-compatible clients like Claude Desktop, Cursor, and Cline AI.18 npmMIT
- AlicenseNot gradedqualityAmaintenanceEnables AI assistants to research topics, plan software, generate requirements, create starter projects, map codebases, curate context, manage tasks, and run multi-step development workflows through MCP-compatible clients.18 npmMIT