BuildBetter Skills: coding skills backed by customer evidence
BuildBetter Skills
Skill packs for Claude Code, Codex and Cursor: specs, plans and browser verification enriched with BuildBetter customer evidence
- 24 ★ on the repository
- Last commit
Extended access
Why this level
- Browser testing skills walk through the running app via Playwright MCP and can change its data
- BUILDBETTER_API_KEY opens your BuildBetter customer data to the skills
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add skills/buildbetter-skillsDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- Browser testing skills walk through the running app via Playwright MCP and can change its data
- BUILDBETTER_API_KEY opens your BuildBetter customer data to the skills
This is third-party code. Review the repository files before installing.
What it does
BuildBetter Skills groups skills into packs and installs only the ones you need. The spec-workflow pack drives development from a specification: bb-specify, bb-plan, bb-tasks, bb-implement, plus bb-clarify, bb-analyze, bb-checklist, bb-constitution and bb-review. The testing pack holds app-navigator, which maps routes and components, trust-but-verify, which checks in a browser that the interface matches the plan, and generate-tests, which produces Playwright test files. The smart-tags pack works with BuildBetter Smart Tags: create, evaluate and improve. The support pack triages customer issues: investigation, resolution, communication. If the BUILDBETTER_API_KEY variable is set, the spec skills pull customer feedback, feature requests and complaints into the spec, plan and review. Without a key they work as usual. Install with npx skills add, through the bb-skills CLI (pip) that detects platforms automatically, or as Codex and Claude Code plugins. Some packs connect the BuildBetter MCP and CLI. The spec skills originate from github/spec-kit.
Who it is for. Product teams developing with Claude Code or Codex who want specs backed by real customer feedback from BuildBetter.
Good fit when
- A new feature starts with a spec that needs real customer requests in place of team guesses
- A branch is ready to merge and you want to check in a browser that the interface matches the plan
- The app needs Playwright test files built from a route map
- A customer issue has to be investigated with evidence and a reply drafted
- A BuildBetter Smart Tag needs evaluation and a bounded backfill over call history
Not a fit when
- You need only a general spec method with no customer data: the original Spec Kit is simpler
- The project does not use BuildBetter: some packs rely on its MCP and CLI
- You need to verify server-side logic only: trust-but-verify checks the interface in a browser and backend-only changes do not fit it
Example request
/bb-specify a user dashboard with an activity feed, using customer feedback from BuildBetterLimitations
Skills adapt to the platform: Claude Code and Codex get SKILL.md with slash commands, while Cursor, Copilot, Windsurf and Amazon Q get passive rules with no commands. Customer evidence is pulled only when BUILDBETTER_API_KEY is set and only as an extra: with no key or an unreachable API, work goes on.
Installing with pip installs the bb-skills CLI, and the Codex package wires in Playwright MCP for the browser testing skills. BuildBetter plugins for Codex and Claude Code are separate packages. Access terms for the BuildBetter platform are not described in the repository.
How to disable. Remove skills with bb-skills uninstall <pack>, and remove skills installed through npx from the agent's skills folder, for example ~/.claude/skills/. Unset the BUILDBETTER_API_KEY variable.
FAQ
How do I install BuildBetter Skills in Claude Code?
Run npx skills add buildbetter-app/skills, or install a pack through the CLI: pip install bb-skills and bb-skills install spec-workflow. For the BuildBetter plugin use claude plugin marketplace add buildbetter-app/skills.
Do I need a BuildBetter key?
No. Without BUILDBETTER_API_KEY the skills work as ordinary ones, and with a key they add customer evidence to specs, plans and reviews.
Which packs exist?
core, spec-workflow, testing, smart-tags and support. Install one with bb-skills install <pack> or everything with bb-skills install all.
Which platforms are supported?
Claude Code, Codex, Cursor, GitHub Copilot, Gemini CLI, Windsurf and Amazon Q. bb-skills platforms shows what is found on the machine.
How does it differ from Spec Kit?
The spec foundation comes from github/spec-kit. BuildBetter added customer evidence, browser testing, support for several platforms and a skill manager.
Discussion
No questions or reviews yet.
Related
SkillsSpec Kit skills for Claude Code, Codex, Cursor, Copilot and OpenCode: specify, plan, tasks, implement, baseline of existing code
A skills library for Claude Code, Cursor and Codex: team code review, TDD, stacked PRs and repository readiness scoring
Two skills for spec-driven development: /spec asks questions and saves a document to specs/, /spec-impl implements an approved spec step by step
speq drives spec-driven development in Claude Code and Codex: mission, plan, implement and record into a permanent library.
61 small skills for Codex across 13 areas: codebase orientation, tests, migrations, security, CI and pull request review
bigpowers gives an agent 80 skills with a six-phase development cycle, quality gates and specs kept in a specs folder
Comparison with similar tools
Cards that share the main search query, and how they differ.
| Tool | Official | Repository stars | Main difference |
|---|---|---|---|
| BuildBetter Skills: coding skills backed by customer evidence (this card) | yes | 24 | |
| Spec Kit | no | 149,434 | the original spec toolkit, which is extended here with customer evidence, browser verification and a skill manager |
| BMad Method | no | 55,288 | a large agent process with roles, whereas BuildBetter Skills ships separate installable packs and pulls customer evidence |
README in short
The README explains three install methods, automatic platform detection, packs with skill lists, a format table per platform, the customer evidence integration and bb-skills CLI commands. The license is stated as MIT.
README badge
Original SKILL.md
---
name: trust-but-verify
description: Use when a feature branch has been implemented and you need to verify the UI/UX and functionality match the original plan before merging
allowed-tools: Bash, Read, Write, Glob, Grep, Agent, mcp__playwright__*
---
# Trust But Verify
Verify that a feature implementation actually matches its plan by testing it in a real browser.
**Core principle:** Plans describe intent. Code describes implementation. Only the browser shows reality. This skill bridges all three — reading the plan, analyzing the diff, and verifying the result in a live browser.
## When to Use
- After completing a feature branch and before merging
- When a plan exists in `docs/plans/` for the current work
- When the diff touches frontend source files (UI changes)
- When you want confidence that the UI matches the spec
- When recommended by `superpowers:finishing-a-development-branch`
**Not for:** Backend-only changes, API-only work, or branches without a plan.
## Process
```dot
digraph trust_but_verify {
"Phase 1:\nGather Context\n(subagent)" [shape=box];
"Phase 2:\nPreflight Check\n(main)" [shape=box];
"Phase 3:\nBrowser Verification\n(main + MCP)" [shape=box];
"Phase 4:\nReport\n(subagent)" [shape=box];
"Print summary\n+ link to report" [shape=doublecircle];
"Phase 1:\nGather Context\n(subagent)" -> "Phase 2:\nPreflight Check\n(main)";
"Phase 2:\nPreflight Check\n(main)" -> "Phase 3:\nBrowser Verification\n(main + MCP)";
"Phase 3:\nBrowser Verification\n(main + MCP)" -> "Phase 4:\nReport\n(subagent)";
"Phase 4:\nReport\n(subagent)" -> "Print summary\n+ link to report";
}
```
### Phase 0: Check for App Navigator
Before starting, check if `~/.claude/skills/app-navigator/app-map.md` exists.
**If it does NOT exist:**
> "I notice the app hasn't been mapped yet. Running `/app-navigator setup` first will map all your routes, build login playbooks, and document UI patterns — which makes verification much faster and more accurate.