skills.rest
PolymathWizardPolymathWizardCommunity·1 Agent Skills Included

BHIL-AI-First-Development-Toolkit

Specification-driven sprint methodology for building LLM-powered applications

Guides solo developers through building LLM-powered applications with a traceable PRD-to-deploy artifact chain. Provides ready-to-fill templates for requirements, specs, architecture decisions, tasks, evaluations, and safety guardrails. Eliminates vague specifications, context loss between sessions, and untested model behavior. Includes scripts and skills that scaffold features, sprints, and decision records automatically.
129|24
↗
npx skills add PolymathWizard/BHIL-AI-First-Development-Toolkit --all -g -y
Available:

Instructs the agent to enforce the PRD-to-SPEC-to-ADR-to-task artifact chain, follow always/ask/never rules, manage context budgets, and use the toolkit's scaffolding commands before writing any code.

All Skills in This Repository (1)

Pure Emerald Level Indicators
📦 In Repo
PolymathWizardPolymathWizard

pdf

Extract text and structured data from PDF documents using pypdf and pdfplumber.

Community
Intermediate

Frequently Asked Questions

FAQPage Schema
How to install BHIL AI-First Development Toolkit?▼

Run `npx skills add PolymathWizard/BHIL-AI-First-Development-Toolkit --all -g -y` in your terminal to install all skills globally.

What is AI-first development?▼

It is a workflow where coding agents implement features while humans focus on specifications, architecture decisions, and review. This toolkit provides the templates and sprint process to make that handoff reliable.

How does the PRD to SPEC to ADR chain work?▼

Every feature starts with an approved requirements document, then a technical spec, then decision records, and finally task breakdowns. Each artifact links to its parents via traceability IDs so nothing is built without approval.

Does this toolkit work with Claude Code and Cursor?▼

Yes. The AGENTS.md file is vendor-neutral and works with Claude Code, Cursor, Devin, GitHub Copilot, and other compatible coding tools.

How do I write acceptance criteria for LLM features?▼

Use probabilistic bands instead of exact outputs, such as a 90% pass rate over 50 runs. The toolkit's templates and guides show exactly how to structure these measurable criteria.

Related Repositories in Software Engineering

View All in Software Engineering→
openclawopenclaw
388.4k

openclaw

Run a personal AI assistant across your devices and chat apps

Official📦 120 Skills
obraobra
280.0k

superpowers

Gives coding agents a disciplined workflow from idea to merged code

Community📦 15 Skills
facebookfacebook
248.4k

react

AI agent skills for building, testing, and porting React core

Official📦 14 Skills