Skill Claude CodeCodex
Guidelines for adopting specialized agent personas, Coordinator Mode, and managing context compression. Activate when performing complex design decisions, delegating to subagents, coordinating multi-file changes, or managing long conversation histories.
Skill Claude CodeCodex
Rules for managing tasks, project documentation, cross-referencing, and cross-session memory. Activate when resolving project tasks, creating documentation, referencing files, or logging persistent session memories.
Skill Claude CodeCodex
Guidelines and optimizations to reduce LLM token usage during development and command execution. Activate when executing shell commands, running large test suites, reading files, searching codebases, or staging commits.
Skill Claude CodeCodex
Guides API design decisions across REST, GraphQL, and tRPC, covering style selection, response envelopes, versioning, pagination, authentication, rate limiting, and security review. Use it when picking an API paradigm for a given audience, shaping consistent payloads and error formats, planning how an interface will…
Skill Claude CodeCodex
Turns plain-language app requests into working full-stack projects. It infers the kind of application being asked for, picks a sensible technology stack, lays out the folder structure, and hands work off to specialist agents. Reach for it when starting a brand-new project, deciding which frameworks to use, or…
Skill Claude CodeCodex
Starter scaffolds for spinning up new applications. Reach for these when bootstrapping a project from nothing. Ships 13 starters spanning common tech stacks.
Skill Claude CodeCodex
A framework for making and recording architecture decisions. Covers gathering requirements and context, weighing trade-offs, picking suitable patterns, and writing decision records (ADRs). Reach for it when deciding how to structure a system, comparing options, capturing the reasoning behind a choice, or reviewing an…
Skill Claude CodeCodex
A working reference for the Bash shell on Linux and macOS. Covers command chaining, file and process handling, text pipelines, scripting scaffolds, and safe error handling. Useful whenever shell work happens on a Unix-like machine instead of Windows.
Skill Claude CodeCodex
A disciplined workflow for applying one repeated change across many files at once — codebase-wide renames, adding imports everywhere, bumping versions, or migrating an API call. It stresses previewing the impact, executing the sweep, and verifying afterward. Reach for it on repetitive multi-file edits, not isolated…
Skill Claude CodeCodex
A catalog of operating modes that reshape how an assistant approaches, communicates, and prioritizes work depending on the task at hand. Covers ideation, building, debugging, reviewing, teaching, releasing, plus multi-agent collaboration patterns. Useful for matching tone and method to whether the work is exploratory…
Skill Claude CodeCodex
A clarify-before-building protocol that turns vague or complex requests into well-scoped work. It defines a questioning gate that pauses implementation to ask decision-focused questions with explicit trade-offs and sensible defaults, then keeps the user informed through progress status boards, plain-language error…
Skill Claude CodeCodex
Generates a Keep a Changelog file and cuts a Semantic Versioning release. Derives the major/minor/patch bump from Conventional Commit history, maintains the Unreleased section, sorts entries into Added/Changed/Deprecated/Removed/Fixed/Security, then tags vX.Y.Z and drafts release notes. Use when preparing a release…
Skill Claude CodeCodex
Pragmatic coding standards favoring concise, direct, well-named code over verbose tutorials, needless comments, or speculative abstraction. Covers core principles, naming, function shape, anti-patterns, pre-edit dependency checks, and a discipline for handling validation-script output. Applies to essentially all…
Skill Claude CodeCodex
A practical checklist for reviewing code across correctness, security, performance, quality, testing, and documentation, plus review patterns specific to AI- and LLM-driven code. It also offers anti-patterns to flag and a tagging convention for review comments. Useful when reviewing a pull request or auditing a change…
Skill Claude CodeCodex
An MCP server that parses a codebase into an AST graph (Tree-sitter into SQLite) so an assistant can fetch only the files in a change's blast radius instead of reading everything. It cuts token use on large repos and supports dead-code detection, refactor previews, and architecture maps. Best on big or multi-repo…
Skill Claude CodeCodex
A technique for keeping long working sessions sharp by condensing finished work while keeping the decisions that still matter. It covers when to compress, three levels of compression from single tool outputs to full checkpoints, and a step-by-step protocol. Useful once a session stretches long, turns get repetitive…
Skill Claude CodeCodex
Manages what enters an agent's context window — instructions, retrieved knowledge, tools, memory, and history — under a token budget. Use when building agents or RAG features, when long context degrades answer quality, when retrieval pulls in noise, or when deciding what to preload vs fetch on demand. Optimizes the…
Skill Claude CodeCodex
Writes commit messages that follow the Conventional Commits 1.0 spec — type(scope): subject with an imperative subject, optional body, and breaking-change markers (! and BREAKING CHANGE:). Use when committing changes, reviewing a commit message, or wiring automated changelogs and semantic versioning. Backs AIM's…
Skill Claude CodeCodex
Orchestrates many worker agents at once, breaking a request into subtasks, running independent reads in parallel, gating writes sequentially, and merging the returns into one answer. Use it when a job spans several domains and benefits from delegated workers plus deliberate synthesis, or when /coordinate or…
Skill Claude CodeCodex
Guides data-layer decisions for application backends, covering schema modeling, index strategy, ORM and database engine selection, query tuning, and migration planning. Applies when shaping a new schema, weighing storage engines or ORMs, planning safe migrations, diagnosing slow queries, or touching Prisma, Drizzle…
Skill Claude CodeCodex
Covers the judgment behind safe production releases — choosing a platform, verifying before shipping, watching a rollout, and deciding when to roll back versus fix forward. It teaches reasoning rather than copy-paste commands, since the right move depends on the stack and the change. Reach for it when planning a…
Skill Claude CodeCodex
Provides starting structures and conventions for the documents a project usually needs — READMEs, per-endpoint API references, doc comments, changelogs, decision records, and machine-readable files for AI agents. Each template is a skeleton to adapt, paired with guidance on what belongs where. Use it when authoring or…
Skill Claude CodeCodex
Writes acceptance criteria in EARS notation (Easy Approach to Requirements Syntax) so each one states a single atomic, testable behavior using the five patterns — ubiquitous, event-driven WHEN, state-driven WHILE, unwanted-behavior IF/THEN, and optional WHERE. Use when authoring a spec's requirements, defining what…
Skill Claude CodeCodex
A reasoning framework for web UI design decisions — components, layout, color, type, effects, and motion. Use it when composing interfaces, picking palettes or fonts, structuring page layouts, or pushing a design past generic defaults. It teaches the principles and trade-offs behind each choice rather than prescribing…