buildcrew

18 AI agents + 7 Agent Skills for Claude Code — adversarial challengers, a spec-compliance gate, and a parallel verification fleet.

$ npx buildcrewclick to copy

Get Started GitHub

Agents

Modes

Opus

Pipeline Docs

How it works

Three steps. Zero config files to write.

Install

npx buildcrew

18 agents + 7 skills copied to your project

Init

npx buildcrew init

Auto-detects your stack, generates harness

Build

@buildcrew [your request]

Orchestrator routes to the right agents

18 Agents

9 opus for critical thinking, 9 sonnet for execution.

Buildopus

planner

6 Forcing Questions + 4-Lens Self-Review

Buildopus

designer

Web research → Playwright → Figma → Production code

Buildopus

developer

6 Implementation Questions + 3-Lens Self-Review

Adversarialopus

plan-challenger

6-vector attack on the plan before design

Adversarialopus

spec-challenger

8-vector attack on the spec before dev

Qualitysonnet

qa-tester

Types, lint, build, acceptance criteria

Qualitysonnet

browser-qa

4-phase browser testing via Playwright MCP

Qualitysonnet

spec-verifier

Spec-compliance gate — every AC verified with file:line evidence

Qualityopus

reviewer

4-specialist parallel review + adversarial + auto-fix

Qualitysonnet

health-checker

3-phase code quality (detect, measure, prescribe)

Sec & Opssonnet

security-auditor

OWASP Top 10 + STRIDE + 10-phase audit

Sec & Opssonnet

canary-monitor

3-phase post-deploy health (orient, verify, judge)

Sec & Opssonnet

shipper

Test → version → changelog → PR

Thinkingopus

thinker

6 forcing questions + premise challenge + design doc

Thinkingopus

architect

Architecture review — scope, diagrams, failure modes

Thinkingsonnet

design-reviewer

8-dimension UX scoring (0-10) + specific fixes

Specialistsonnet

investigator

Root cause debugging, edit freeze

Specialistopus

qa-auditor

3 parallel subagent audit on git diffs

Specialistopus

coherence-auditor

Verifies the team actually coordinated — score 0-100%

View all agents →

Feature Pipeline

Every feature generates a full document chain.

Plan

01-plan.md

→

01.5

Challenge

01.5-plan-critique.md

→

Design

02-design.md

→

02.5

Challenge

02.5-spec-critique.md

→

Dev

03-dev-notes.md

→

04-qa-report.md

→

Browser

05-browser-qa.md

→

03.5

Spec-Verify

03.5-spec-verification.md

→

Review

06-review.md

→

Ship

07-ship.md

Learn more →

13 Operating Modes

Talk naturally. Orchestrator auto-detects the mode.

🏗️

Feature

@buildcrew Add user dashboard

Plan → Design → Dev → [QA ∥ Browser ∥ Spec-Verify] → Review

🔍

Audit

@buildcrew full project audit

Scan → Prioritize → Fix → Verify

🌐

Browser QA

@buildcrew browser qa localhost:3000

Playwright tests + health score

🛡️

Security

@buildcrew security audit

OWASP + STRIDE + secrets + deps

🔬

Debug

@buildcrew debug: login broken

4-phase root cause investigation

📊

Health

@buildcrew health check

Quality dashboard (0-10 score)

🐤

Canary

@buildcrew canary https://myapp.com

Post-deploy monitoring

👁️

Review

@buildcrew code review

Multi-specialist + auto-fix

🚀

Ship

@buildcrew ship

Test → version → changelog → PR

🔍

QA Audit

@buildcrew qa

3 parallel subagent audit on git diffs

💭

Think

@buildcrew think: should we add this?

6 forcing questions + alternatives + design doc

🏛️

Arch Review

@buildcrew architecture review

Scope + diagrams + failure modes + test map

🎨

Design Review

@buildcrew design review

8-dimension scoring + screenshot evidence

View all modes →

Harness Engineering

Your project context baked into every agent. Zero questions asked.

▌

Learn more →

Ready to build?

$ npx buildcrewclick to copy

Read the Docs npm