andrewtliem/ai-native-app-builder-skills
13 skills13 installs26 starsGitHub
Install
npx skills add https://github.com/andrewtliem/ai-native-app-builder-skillsSkills in this repo
1Agent RulesAgent Rules is a Claude skill that generates project AI rule files such as AGENTS.md or .cursorrules. A developer uses it after the PRD and architecture are stable so AI coding tools operate inside the project's intent, stack, coding conventions, and verification expectations. It defines allowed and forbidden changes, the exact install/run/test/build commands, and when the AI must ask the human instead of guessing.1installs2Architecture To IssuesArchitecture to Issues is a Claude skill that turns a PRD and architecture plan into a backlog of small, ordered implementation issues. A developer uses it before coding to split work into reviewable units, each with a goal, dependencies, likely files, acceptance criteria, and verification steps. It marks what AI may implement versus what the human must judge, avoiding giant issues like 'build the whole app'.1installs3Demo ReadyDemo Ready is a Claude skill that prepares an app for presentation. A developer uses it after the app passes basic tests to produce a demo script with a problem statement, a two-to-five-minute demo flow, completed features, evidence, and honest limitations. It also records what AI helped with and what the human decided, and ends with next-improvement steps.1installs4Grill My IdeaGrill My Idea is a Claude skill that pressure-tests a raw app idea before a developer writes a PRD or asks AI to code. It restates the idea, identifies the likely user and pain, and asks 8-12 hard questions grouped by user, problem, evidence, scope, risk, and success. It separates facts from assumptions, finds the smallest useful version, and ends with a proceed/revise/research decision gate.1installs5Idea To IntentIdea to Intent is a Claude skill that converts a pressure-tested app idea into a short Intent Brief. A developer uses it after grilling the idea, once they can explain the user and problem but have not yet written requirements. It captures the project's purpose, primary user, goals and non-goals, constraints, tradeoffs, and checkable success criteria, and records open questions rather than hiding them.1installs6Intent To PrdIntent to PRD is a Claude skill that turns an Intent Brief into a Product Requirements Document. A developer uses it after the intent is captured and before architecture or issue planning. It writes user stories, core flows, functional and non-functional requirements, and acceptance criteria, while marking out-of-scope features and preserving the original intent rather than inventing a different product.1installs7Issue To PromptIssue to Prompt is a Claude skill that converts a single implementation issue into a focused AI coding prompt. A developer uses it when ready to work one issue with an AI coding tool. The prompt includes only relevant PRD and architecture context, the files to inspect first, constraints and forbidden changes, the exact expected output, and the checks to run, and it requires the AI to explain assumptions and report verification.1installs8PostmortemPostmortem is a Claude skill for reflecting after a project or milestone. A developer uses it to compare original intent to the final result, record where AI accelerated work and where it caused bugs or overconfidence, and note what was and was not verified. It extracts concrete rules to add to future prompts or AGENTS.md and ends with a biggest lesson and one next improvement.1installs9Prd To ArchitecturePRD to Architecture is a Claude skill that turns a PRD into a simple architecture plan before coding. A developer uses it after the PRD and before issue breakdown to define system parts, data flow, data models, key routes or screens, and architecture decisions with reasons. It names security and privacy concerns and failure cases while choosing the simplest stack that satisfies the PRD.1installs10Red Team My AppA Claude skill for students that reviews an app for misuse, privacy, security, reliability, accessibility, and UX risks before a demo or submission. It reads the project PRD, architecture, loop log, and test plan, then produces a prioritized Red Team Report artifact. A developer uses it to catch authentication, input-validation, upload, data-retention, and prompt-injection risks and separate class-demo risk from production risk.1installs11Review The CodeA Claude skill that reviews student or AI-generated code against the issue, PRD, architecture, readability, security, and privacy. It compares the implementation to acceptance criteria, checks for unrelated changes and missing verification, and tests whether the student can explain the code. A developer uses it before merging or moving to the next issue to keep AI-written code aligned with intent.1installs12Run The LoopA Claude skill that walks a student through completing one implementation issue in a controlled AI-assisted loop: issue, prompt, build, review, test, fix, decide, repeat or accept. It caps the loop at one to three cycles, records evidence in a Loop Log, and keeps the developer responsible for intent and final judgment. A developer uses it to get a repeatable build rhythm instead of an endless AI coding session.1installs13Test The AppA Claude skill that converts requirements and implemented features into a practical test plan with manual checks, acceptance tests, edge cases, and demo verification. It maps each PRD acceptance criterion to a test, adds failure and misuse cases, and lists evidence to collect such as screenshots, logs, and a deployed URL. A developer uses it before a demo or submission to prove the app works rather than trusting that the AI said it is done.1installs