BuildBetter
Research customer feedback, calls, support, surveys and product usage with source citations.
Install
Add this to your client's MCP configuration (Claude Desktop: claude_desktop_config.json; Cursor: .cursor/mcp.json; VS Code and others use the same "mcpServers" shape).
claude mcp add --transport http buildbetter 'https://mcp.buildbetter.app'
{
"mcpServers": {
"buildbetter": {
"type": "http",
"url": "https://mcp.buildbetter.app"
}
}
}
Documentation
From the project's README on GitHub, refreshed weekly. Links open on GitHub.
Contents
BuildBetter Skills
AI coding skills enriched with BuildBetter customer evidence.
What is this?
BuildBetter Skills is a collection of AI coding skills you can install into your favorite AI coding agent. Pick the skills you need -- install only the packs that matter for your workflow. BuildBetter Skills works with Claude Code, Codex, Cursor, GitHub Copilot, Gemini CLI, Windsurf, and Amazon Q.
Quick start
Option A: npx (fastest, no install)
npx skills add buildbetter-app/skills
Or install a specific skill:
npx skills add buildbetter-app/skills --skill trust-but-verify
Option B: bb-skills CLI (pack-based install)
pip install bb-skills
bb-skills install spec-workflow # spec-driven development skills
bb-skills install testing # browser testing skills
bb-skills install all # everything
Option C: Codex plugin package
This repo also includes a Codex plugin package at plugins/skills/ plus a repo-local marketplace entry at .agents/plugins/marketplace.json. The plugin bundles the same BuildBetter Skills prompts for Codex and includes Playwright MCP wiring for the browser-testing skills.
codex plugin marketplace add buildbetter-app/skills --ref main --sparse .agents/plugins --sparse plugins/skills --sparse plugins/buildbetter-codex
codex plugin add skills@buildbetter
Option D: BuildBetter MCP/CLI plugins
This repo includes separate BuildBetter MCP/CLI plugins for Codex and Claude Code under plugins/buildbetter-codex/ and plugins/buildbetter-claude/. They include BuildBetter-specific skills for MCP routing, customer voice, synthetic research, native surveys, Smart Tags, project triage, and knowledge gaps. They are separate packages because Codex and Claude Code use different plugin manifest directories and remote MCP config schemas.
Codex:
codex plugin marketplace add buildbetter-app/skills --ref main --sparse .agents/plugins --sparse plugins/skills --sparse plugins/buildbetter-codex
codex plugin add buildbetter@buildbetter
Claude Code:
claude plugin marketplace add buildbetter-app/skills --sparse .claude-plugin plugins/buildbetter-claude
claude plugin install buildbetter@buildbetter
Option E: Grok, Cursor and MCP marketplaces
The existing Grok/Cursor plugin uses BuildBetter's
production MCP and nine public research workflows. For the MCP Registry, Google
Connected Apps, Meta, Microsoft 365, Salesforce and AWS preparation, see the
distribution guide. Build a reproducible kit with
python3 scripts/build_mcp_distribution.py. Provider registration, native-host
QA and marketplace approval are separate from building the package.
3. Use them
Skills are installed into your AI coding agent. On platforms with slash commands (Claude Code, Codex, Gemini):
/bb-specify Build a user dashboard with activity feed
/bb-plan
/trust-but-verify
On passive platforms (Cursor, Copilot, Windsurf, Amazon Q), skills are injected as context automatically.
How it works
You run bb-skills install once per machine. The CLI auto-detects which AI coding platforms you have installed (~/.claude/, ~/.codex/, ~/.gemini/, .cursor/, etc.) and copies skills to all of them in the correct format.
To see what platforms are detected:
bb-skills platforms
To install for a specific platform only:
bb-skills install testing --platform claude
To install a single skill instead of a whole pack:
bb-skills install trust-but-verify
Available packs
| Pack | Skills | Description |
|---|---|---|
core | bb-skills-update | Update checker (auto-included) |
spec-workflow | bb-specify, bb-plan, bb-review, bb-tasks, bb-clarify, bb-analyze, bb-checklist, bb-constitution, bb-implement | BuildBetter-enriched spec-driven development |
testing | app-navigator, trust-but-verify, generate-tests | Browser-based verification and Playwright test generation |
Platform support
| Platform | Format | Slash Commands | Install Path |
|---|---|---|---|
| Claude Code | SKILL.md | Yes | ~/.claude/skills/ |
| Codex CLI | SKILL.md | Yes | ~/.codex/skills/ |
| Cursor | .mdc rules | No (passive) | .cursor/rules/ |
| GitHub Copilot | .instructions.md | No (passive) | .github/instructions/ |
| Gemini CLI | .toml | Yes | ~/.gemini/commands/bb-skills/ |
| Windsurf | .md rules | No (passive) | .windsurf/rules/ |
| Amazon Q | .md rules | No (passive) | .amazonq/rules/ |
Skill reference
core
| Skill | Description |
|---|---|
| bb-skills-update | Checks for available BuildBetter Skills updates via GitHub releases |
spec-workflow
| Skill | Description |
|---|---|
| bb-specify | Define what you want to build -- requirements and user stories |
| bb-plan | Create a technical implementation plan with your chosen tech stack |
| bb-review | BuildBetter-first usability and code review |
| bb-tasks | Generate an actionable task list from your implementation plan |
| bb-clarify | Clarify underspecified areas before planning |
| bb-analyze | Cross-artifact consistency and coverage analysis |
| bb-checklist | Generate quality checklists that validate requirements completeness |
| bb-constitution | Create or update project governing principles |
| bb-implement | Execute all tasks to build the feature according to the plan |
testing
| Skill | Description |
|---|---|
| app-navigator | Map application routes, pages, and components for UI/UX audit |
| trust-but-verify | Verify UI/UX and functionality match the original plan |
| generate-tests | Generate reusable Playwright test files from the app map |
BuildBetter integration
The spec-workflow skills can optionally use BuildBetter's customer evidence API to enrich specifications, plans, and reviews with real customer data. This is not required -- all skills work without it. When a BUILDBETTER_API_KEY environment variable is set, the workflow will automatically pull relevant customer insights, feature requests, and pain points into your specs and plans.
To enable it, set the API key in your shell:
export BUILDBETTER_API_KEY=your_key_here
Evidence collection is best-effort and never blocks the workflow. If the key is missing or the API is unreachable, skills continue normally.
Updating
To update all installed skills to the latest version:
bb-skills update
CLI reference
| Command | Description |
|---|---|
bb-skills install <pack> | Install a skill pack |
bb-skills uninstall <pack> | Remove an installed skill pack |
bb-skills update | Update all installed packs to latest |
bb-skills list | List installed packs and skills |
bb-skills platforms | Show supported platforms and install paths |
Contributing
See CONTRIBUTING.md for guidelines on adding skills, writing adapters, and submitting pull requests.
Acknowledgments
BuildBetter Skills was inspired by and originally derived from github/spec-kit (MIT License). BuildBetter has significantly extended the original with customer evidence integration, browser testing skills, multi-platform support, and a skill management CLI.
License
This project is licensed under the MIT License. See LICENSE for details.
Frequently asked questions
What is BuildBetter?
Research customer feedback, calls, support, surveys and product usage with source citations.
How do I install BuildBetter?
It runs as a remote server. In Claude Code: claude mcp add --transport http buildbetter 'https://mcp.buildbetter.app'. For other clients, paste the JSON configuration from this page into their MCP settings.
Does BuildBetter need an API key?
Its registry entry lists no required key or secret. Some features may still need an account with the service it connects to.
Is BuildBetter open source?
Its code is public on GitHub, but the repository does not declare a standard license.
Who maintains BuildBetter?
It is published by buildbetter-app. AI.tn lists it but does not maintain or endorse it.
From the official MCP Registry. Listing is not an endorsement: review a server's code and permissions before giving it access to your data.