gstack-lite

Portable, low-token methodology skills distilled from the public GStack methodology.

What you get

Seven Markdown skills for AI-assisted work:

• shape rough ideas before a PRD exists;
• review product scope and strategic fit;
• review implementation plans before coding;
• explore multiple visual directions;
• create static HTML references, lightweight clickable prototypes, or handoff artifacts;
• review UI/UX readiness;
• run evidence-based QA on built features or demos.

Use them as Claude Code skills or as plain prompts in any capable agent.

What this is not

This package is not GStack. It does not include the GStack runtime, browser daemon, telemetry, dashboards, local ~/.gstack state, update checks, proactive routing, or auto-ship behavior.

The goal is to keep the useful review and artifact-generation discipline in small Markdown skills that can travel between projects.

Included skills

Skill	Use for	Core question
gstack-lite-office-hours	early idea shaping before a PRD or plan exists	What is the real problem, user/operator, evidence, and narrowest useful wedge?
gstack-lite-ceo-review	PRDs, feature plans, roadmap/scope decisions	Are we solving the right problem at the right scope?
gstack-lite-eng-review	implementation plans, architecture, build readiness	Is this plan buildable, testable, and safe to implement?
gstack-lite-design-shotgun	standalone visual exploration, direction briefs, and generation prompts	Which distinct visual directions are grounded enough to choose or refine?
gstack-lite-design-html	standalone static HTML reference, lightweight clickable prototype, or implementation handoff	What should the reference/prototype express, and what is not production?
gstack-lite-design-review	standalone UI/UX plan or implemented-screen audit	Is the user experience specified and intentional enough to build or ship?
gstack-lite-qa-review	built features, demos, release checks	Did the critical user journeys actually work when inspected or tested?

Each skill is independently invocable. Suggested sequences are examples, not required pipelines. Skills may recommend each other as optional next steps when their own output reveals a need, but no skill depends on another by default.

Current limitation: this pack does not yet include lite code review, debug investigation, or release check. For pre-merge and release workflows, use the current skills as planning/design/engineering/QA gates until those skills are added.

Quick use

1. Open the relevant skills/<skill-name>/SKILL.md.
2. Give your agent that skill plus the relevant PRD, plan, design brief, screenshot, prototype, demo URL, test notes, or diff.
3. Ask for the skill's output format.

Example:

Use `gstack-lite-eng-review` to review this implementation plan.
Mark repo-dependent claims as not verified.
Do not edit files.

Install

Personal Claude Code install

Copy each skill directory into your personal skills folder:

mkdir -p ~/.claude/skills
cp -R skills/gstack-lite-* ~/.claude/skills/

Project Claude Code install

Copy into the project:

mkdir -p .claude/skills
cp -R skills/gstack-lite-* .claude/skills/
git add .claude/skills
git commit -m "chore: add gstack-lite review skills"

Generic prompt usage

Use the SKILL.md files as review or artifact-generation prompts. They are written to run without any GStack binaries or state.

Usage pattern

1. Give the agent the relevant authority files: PRD, plan, architecture notes, design brief, screenshots, demo URL, test notes, or diff.
2. Invoke the appropriate lite skill directly.
3. Use the output format defined inside that skill.
4. Treat BLOCKED or DO NOT SHIP as a real stop, not a vibe.

Design usage guidance

• gstack-lite-design-shotgun is for standalone visual exploration: variant briefs, design hypotheses, generation prompts, shortlist/synthesis, and handoff notes.
• gstack-lite-design-html is for standalone static reference/prototype/handoff generation: it preserves authority, scopes visual references, and separates prototype/reference output from production implementation.
• gstack-lite-design-review is for standalone audit/readiness review: it critiques plans, screenshots, prototypes, or implemented screens.

These design skills may compose, but do not depend on one another. For example, a selected shotgun direction can feed an HTML reference, and an HTML reference can later be reviewed, but either skill can also be used directly from authority docs, screenshots, or user notes.

Optional workflow examples

For a new idea:

1. gstack-lite-office-hours
2. gstack-lite-ceo-review
3. gstack-lite-design-shotgun if visual direction is unresolved
4. gstack-lite-design-html if a static reference/prototype/handoff artifact is useful
5. gstack-lite-design-review if UI/UX readiness needs critique
6. gstack-lite-eng-review
7. Implement using your chosen harness
8. gstack-lite-qa-review

For an already-written plan:

1. gstack-lite-ceo-review if scope or problem fit is still uncertain
2. gstack-lite-design-shotgun, gstack-lite-design-html, or gstack-lite-design-review only when the design problem calls for that specific action
3. gstack-lite-eng-review before implementation

For an already-built feature:

1. gstack-lite-qa-review
2. optionally future gstack-lite-code-review when implemented

Repo map

• skills/ — the portable skills. Each skill is self-contained and carries its own output contract in SKILL.md.
• examples/ — example outputs.
• DISTILLATION_STANDARD.md — durable quality standard for skills.
• SKILL_WORKFLOW.md — maintainer workflow for drafting, revising, and smoke-testing skills.
• SOURCE_MAP.md — provenance map from original GStack methodology to this lite pack.
• NOTICE.md — attribution notice.

If a future skill needs a long reusable template or reference file, keep it inside that skill's own directory and reference it from that skill's SKILL.md.

Maintainer workflow

Use SKILL_WORKFLOW.md when drafting a new skill, revising an existing skill, or smoke-testing whether a skill can run end-to-end as a standalone prompt. Keep durable standards in DISTILLATION_STANDARD.md; keep the practical drafting/testing procedure in SKILL_WORKFLOW.md.

Quality boundary

These skills review, specify, or generate handoff/reference artifacts by default. They do not modify files unless the user explicitly asks. That is intentional: the lite pack is a portable methodology layer, not an autonomous build system.

Each skill should preserve operational pressure through evidence ledgers, alternatives, decision gates, failure maps, verdict criteria, and explicit NOT-in-scope handling. If a skill becomes a generic checklist, it has failed the distillation standard.