superpowers
claudecodexCore skills library for Claude Code: TDD, debugging, collaboration patterns, and proven techniques
15 skills · hooks
AI Software Factory: Safe Factory delivery and Improve Factory maintenance skills, delegate roles, the Stop check, the update check and the leftover doctor.
Plugin safe-factory from marketplace ai-software-factory in seifsg/ai-software-factory. Commands read from the repo's manifest — 7 days ago.
analyze-startAnalyzes a feature, defect, refactor or PR before any coding: reads the actual code, explains purpose and success, maps every consumer and draws ASCII flow charts. Use when work starts on existing or new code, or when asked to analyze. Analysis only.
database-schemaDesigns or revises persistence from the whole product's journeys, invariants and query workloads: compares three designs, writes the schema, plans indexes and migrations, and audits the result. Use when adding or materially changing tables, collections, indexes or migrations. Routine fixes stay scop
deep-reviewDeep code analysis and review that hunts for bugs, oversights, anti-patterns, performance issues, stupid mistakes, potential blockers, wrong implementations, and bad code. Use when wanting a thorough code review, quality audit, or pre-merge verification of implementation quality.
get-ready-to-codePlans the build from a finished analysis and recommendation: files to change, self-critique, phases, the full todo list and the whole plan, then runs every todo through the 0-4 loop. The todo list becomes each ticket's checklist. Use after analysis and before coding. No coding during planning.
how-to-make-a-prPrepares and opens a draft pull request for human review, linked to its ticket and backed by before/after proof. Use when creating, updating or correcting a pull request.
how-to-review-a-prReviews a pull request end to end on a separate, fresh agent: intake, analysis, three solutions, the factory run audit, every review lens, synthesis, bounded fixes, gate and a detailed PR comment. Use for every factory-made draft PR and when asked to review any PR. Never marks ready or merges.
improve-factoryRepairs or improves the factory's own skills, hooks and adapters when factory instructions or behavior fail, in an isolated factory checkout, while any app run keeps going. Use when the factory misbehaved or needs a change. Product bugs go back to delivery.
recommendCompares three strong solutions for a change in one table and recommends exactly one, with reasons: correctness, simplicity, performance, fit to the original goal, coupling. Use after analysis and before planning, when reviewing a PR, or whenever a substantive choice comes up. No coding.
research-marketResearches competitors, products, UX, stacks, architecture and design patterns. Use when starting a new product or big feature, when a material unknown blocks a decision, or when asked for standalone research. Skip for routine fixes.
safe-factoryDelivers games, apps, services, tools and libraries end to end in ten visible steps: ticket, analyze, recommend, challenge, plan, Before proof, build, verify, PR and review. Fast by default, Safe on request. Use when asked to build, extend, fix or resume software with the factory.
stop-checkArms a factory run's Stop check. Only in a run, before a block or todo line.
ticketingCreates and maintains the project backlog, its epics and tickets in every mode. Use when work needs tickets or backlog upkeep, when starting a product, feature or fix, or when a PR or review must be checked against a ticket.
analystFactory delegate: use only when a factory brief names this role. Standard tier, no edits. Researches from primary sources, analyzes one area, reviews through one lens, writes UX verdicts, walks through a changed method read-only or scouts a large spec into one brief per job. Use for judgement that c
builderFactory delegate: use only when a factory brief names this role. Deep tier, edits. Implements a planned change that needs design calls inside its constraints: architecture, schema, security, unclear bugs. Use when a wrong call is expensive. Not for spelled-out edits (maker) or review (reviewer).
checkerFactory delegate: use only when a factory brief names this role. Light tier, no edits. Checks a run's step outputs against their skills' output contracts at gates A, B and C, from their checklist lines only. Not for judging quality (reviewer) or fixing (maker).
makerFactory delegate: use only when a factory brief names this role. Standard tier, edits. Makes planned or mechanical edits a brief spells out: a todo with its plan, fix rounds that name each fix, walkthroughs of a changed method. Use when the change needs no design call. Not for architecture, schema,
reviewerFactory delegate: use only when a factory brief names this role. Deep tier, no edits. Fresh-context final and independent review: the requirements, plan, diff and raw evidence in, ranked findings and a verdict out, never the builder's reasoning. Also deep judgement that changes no file: an unclear b
runnerFactory delegate: use only when a factory brief names this role. Light tier, no edits. Runs a brief's scripted commands with a yes/no result (tests, builds, the full suite), searches, inventories, comparisons and log reads exactly as written. Use when the answer is command output or a list of places
Core skills library for Claude Code: TDD, debugging, collaboration patterns, and proven techniques
15 skills · hooks
Backend API design, GraphQL architecture, workflow orchestration with Temporal, and test-driven backend development
9 skills · 1 command · 8 agents
AI-supervised issue tracker for coding workflows
1 skill · 1 agent · hooks
MiniMax AI skills library for frontend, fullstack, Android, iOS, shader, GIF sticker, document, presentation, spreadsheet, and multimodal media workflows
18 skills
Worktrunk is a CLI for Git worktree management, designed for parallel AI agent workflows. This plugin provides configuration guidance (LLM commit messages, project hooks, worktree paths), automatic activity tracking (🤖/💬 indicators in `wt list` showing active Claude sessions), and a `/wt-diff` pane showing each worktree's changes.
2 skills · hooks
GSD Core is a meta-prompting, context engineering, and spec-driven development system for AI coding agents.
72 skills · 72 commands · 64 agents · hooks
This page updates itself from your repo. Add the badge to your README, or get it Featured to pin it above the organic results with a followed link to your site.