Runs the project's declared quality tooling in three ordered buckets — format, lint, typecheck — re-staging files where appropriate and fail-fast on any hard-stop. Spawned by /implement between programmer and reviewer. Returns per-bucket status, files re-staged, and hard-stop details if any.
Executes a single RED → GREEN → REFACTOR cycle for a described behavior — writes failing tests, makes them pass with the minimum implementation, then refactors cleanly. Spawned by /implement for each SUBTASK. Returns files changed, test names added, and RED→GREEN evidence.
Reviews staged or specified code changes against a fixed core checklist (TDD, SOLID, DRY, naming, deep modules) plus adaptive extras based on the type of change. Spawned by /implement after linter to produce a PASS/FAIL verdict. Returns BLOCKING findings, ADVISORY findings, and the verdict.
Drafts a Nygard-format ADR or a style guide entry from a TASK issue, presents it for human confirmation, writes it to the appropriate docs/ directory, and handles supersession. TRIGGER when: user says 'document this decision', 'write an ADR', 'draft an ADR', 'create an ADR', 'write a style guide entry', or /implement is working on an ADR-labeled or style-guide-labeled TASK.
Breaks a PRD into vertical-slice TASKs by grouping user stories that share the same end-to-end context, proposes the breakdown with rationale for human approval, then creates the approved TASKs as child GitHub issues. TRIGGER when: user says 'decompose the PRD', 'break into tasks', 'create tasks for issue #N', 'split this into tasks'.
Orchestrates the full TDD implementation loop for an ai-ready TASK — maps acceptance criteria to SUBTASKs, spawns the programmer/linter/reviewer subagents per SUBTASK, commits atomically, runs the full suite, then opens a PR after human approval. TRIGGER when: user says 'implement task #N', 'start implementing', 'work on this task' or 'implement this'.
Conducts a relentless requirements interview — one question at a time with recommended answers and pushback — until a full shared understanding is confirmed and handed off to /prd. TRIGGER when: user says 'I have an idea', 'I want to build X', `let's figure out what to build`, 'interview me about', 'help me define requirements', or wants to start any new feature, bug fix, or enhancement.
Manages GitHub issues for the software factory — creates, reads, updates, and lists PRD and TASK issues; transitions states via labels; shows a status dashboard; and recovers missing labels on demand during write operations. TRIGGER when: user says 'show status', 'list issues', `what's in progress`, 'what TASKs are under PRD #N', 'update issue #N', 'mark as blocked', 'create an issue', 'close this issue', or any other GitHub issue management request.
Uses power tools
Uses Bash, Write, or Edit tools
Own this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimOwn this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimBased on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
A Claude Code plugin that turns AI-assisted development into a repeatable software factory, taking you from a vague idea to a merged PR through interview → PRD → decomposition → TDD implementation → review.
I started this collection while figuring out how to leverage AI to actually ship software. I was hesitant for a long time, but once I started using it the workflow turned out to be pretty neat — these skills are what I ended up with.
gh CLI, authenticatedclaude plugin marketplace add fslzrr/skills
claude plugin install fslzrr@fslzrr-marketplace
/plugin marketplace add fslzrr/skills
/plugin install fslzrr@fslzrr-marketplace
| Stage | Skill | Description |
|---|---|---|
| Define | /interview | Conducts a relentless requirements interview — one question at a time with recommended answers and pushback — until a full shared understanding is confirmed and handed off to /prd. |
| Define | /prd | Explores the codebase deeply, then structures confirmed /interview findings into a PRD GitHub issue — mapping interview narrative to template sections, identifying modules to build or modify, and surfacing deepening opportunities. |
| Plan | /decompose | Breaks a PRD into vertical-slice TASKs by grouping user stories that share the same end-to-end context, proposes the breakdown with rationale for human approval, then creates the approved TASKs as child GitHub issues. |
| Build | /implement | Orchestrates the full TDD implementation loop for an ai-ready TASK — maps acceptance criteria to SUBTASKs, spawns the programmer/linter/reviewer subagents per SUBTASK, commits atomically, runs the full suite, then opens a PR after human approval. |
| Build | /tdd | Executes a single RED/GREEN/REFACTOR cycle for a described behavior — writes failing tests first, makes them pass with minimal code, then refactors cleanly. |
| Build | /lint | Runs three ordered quality-gate buckets — format (fix + re-stage), lint (check-only → auto-fix → hard-stop), typecheck (project-wide → hard-stop) — each discovered independently from declared project scripts; notifies and skips missing buckets; fail-fast on hard-stop. |
| Build | /review | Reviews staged or specified code changes against a fixed core checklist (TDD, SOLID, DRY, naming, deep modules) plus adaptive extras based on the type of change, and produces a PASS/FAIL verdict with blocking and advisory findings. |
| Document | /document | Drafts a Nygard-format ADR or a style guide entry from a TASK issue, presents it for human confirmation, writes it to the appropriate docs/ directory, and handles supersession. |
| Ops | /issues | Manages GitHub issues for the software factory — creates, reads, updates, and lists PRD and TASK issues; transitions states via labels; shows a status dashboard; and recovers missing labels on demand during write operations. |
The programmer, linter, and reviewer agents are subagents spawned by /implement during the TDD loop. They are not invoked directly — listing them here is so you know what's running inside /implement, not as commands to type yourself.
programmer — Executes a single RED → GREEN → REFACTOR cycle for a described behavior — writes failing tests, makes them pass with the minimum implementation, then refactors cleanly.linter — Runs the project's declared quality tooling in three ordered buckets — format, lint, typecheck — re-staging files where appropriate and fail-fast on any hard-stop.reviewer — Reviews staged or specified code changes against a fixed core checklist (TDD, SOLID, DRY, naming, deep modules) plus adaptive extras based on the type of change.The factory runs as a loop. Each step hands its output to the next.
/interview and answer questions until the spec is locked./prd, which opens a GitHub issue./decompose <PRD#> to split the PRD into ai-ready TASK issues./implement <TASK#> to run the TDD loop and open a PR.Use /issues at any point to see the factory's current state.
npx claudepluginhub fslzrr/skills --plugin fslzrrUltra-compressed communication mode. Cuts ~75% of tokens while keeping full technical accuracy by speaking like a caveman.
Comprehensive UI/UX design plugin for mobile (iOS, Android, React Native) and web applications with design systems, accessibility, and modern patterns
Multi-model consensus engine integrating OpenAI Codex CLI, Gemini CLI, and Claude CLI for collaborative code review and problem-solving.
Curate auto-memory, promote learnings to CLAUDE.md and rules, extract proven patterns into reusable skills.