hamzaPixl

65 mods across 1 repository, 2 stars between them.

admin-dashboard

25

hamzaPixl/pixl-ai

Skill Claude CodeCodex

Build an admin dashboard with data tables, forms, RBAC UI, and charts. Orchestrates Next.js scaffolding with backend API integration. Use when asked to create an admin panel, back-office, or management interface.

2 4mo ago A 48 tokens original MIT

agent-browser

26

hamzaPixl/pixl-ai

Skill Claude CodeCodex

Browser automation for AI agents using the agent-browser CLI (Vercel). Use when you need to navigate pages, fill forms, click elements, take screenshots, verify UI behavior, or run end-to-end verification of a running web app.

2 4mo ago A 51 tokens original MIT

agent-development

27

hamzaPixl/pixl-ai

Skill Claude CodeCodex

Create subagent definitions for Claude Code plugins. Use when writing agent frontmatter, designing triggering descriptions with examples, or defining system prompts and tool access.

2 4mo ago A 33 tokens original MIT

api-audit

28

hamzaPixl/pixl-ai

Skill Claude CodeCodex

Audit API endpoints for naming consistency, response shape uniformity, validation gaps, auth holes, and REST convention compliance. Supports Fastify, Express, FastAPI, and Next.js API routes. Use when asked to audit an API, review endpoint design, check API consistency, or find auth gaps.

2 4mo ago A 63 tokens original MIT

batch

29

hamzaPixl/pixl-ai

Skill Claude CodeCodex

Decompose a transformation into independent units, execute each in a worktree-isolated agent, merge results or create per-unit PRs. Use when applying the same change across many files/modules, running parallel migrations, or batch-processing independent tasks.

2 4mo ago A 50 tokens original MIT

benchmark

30

hamzaPixl/pixl-ai

Skill Claude CodeCodex

Compare local code patterns against real-world open-source implementations from GitHub. Use when you want to see how other projects implement the same pattern (auth, DDD, i18n) or identify gaps vs industry standards. Not for performance benchmarking — see /eval-harness for metric-based evaluation.

2 4mo ago A 61 tokens original MIT

blog

31

hamzaPixl/pixl-ai

Skill Claude CodeCodex

Build a blog with MDX content, listing pages, RSS feed, and SEO. Uses the Next.js studio stack. Use when asked to create a blog, add a blog to an existing site, or build a content-driven site.

2 4mo ago A 49 tokens original MIT

cartographer

32

hamzaPixl/pixl-ai

Skill Claude CodeCodex

Analyze a pull request diff and group changes into semantic feature clusters. Use when breaking down a PR into reviewable components, quantifying churn, or explaining change boundaries.

2 4mo ago A 36 tokens original MIT

caveman-commit

33

hamzaPixl/pixl-ai

Skill Claude CodeCodex

Generate ultra-terse Conventional Commits messages. Subject ≤50 chars, body only when the "why" is non-obvious. Preserves intent and reasoning while cutting noise. Use when the user says "write a commit message the caveman way", "terse commit", or invokes /caveman-commit. Does NOT replace /commit-commands:commit …

2 4mo ago A 91 tokens copy · 83% MIT

caveman-compress

34

hamzaPixl/pixl-ai

Skill Claude CodeCodex

Compress natural-language memory files (CLAUDE.md, rules, references, todos) into caveman-speak to save input tokens on every session. Preserves code, URLs, file paths, commands, markdown structure, and headings exactly. Original saved as FILE.original.md for rollback. Uses a vendored Python CLI that calls Claude to…

2 4mo ago A 118 tokens original MIT

caveman-help

35

hamzaPixl/pixl-ai

Skill Claude CodeCodex

Show a quick-reference card listing all caveman modes, skills, and commands. One-shot display, not a persistent mode change. Use when the user says "caveman help", "what caveman commands", "how do I use caveman", or invokes /caveman-help.

2 4mo ago A 64 tokens original MIT

caveman-review

36

hamzaPixl/pixl-ai

Skill Claude CodeCodex

Write ultra-terse PR review comments — one line per finding in the form "L42: 🔴 bug: user null. Add guard." Preserves actionable signal, drops throat-clearing and hedging. Use when the user says "terse code review", "one-line review comments", or invokes /caveman-review. Does NOT replace /code-review — this is a…

2 4mo ago A 90 tokens original MIT

caveman

37

hamzaPixl/pixl-ai

Skill Claude CodeCodex

Switch response style to token-compressed "caveman" mode — drops articles, filler, pleasantries, and hedging while preserving full technical accuracy. Cuts output tokens 30-75% depending on level. Supports intensity levels: lite (default, keeps grammar), full (fragments OK), ultra (extreme compression), wenyan-lite…

2 4mo ago A 121 tokens original MIT

changelog

38

hamzaPixl/pixl-ai

Skill Claude CodeCodex

Generate release notes from conventional commits. Parses commits since the last git tag, groups by type (feat/fix/refactor/etc.), and produces a markdown changelog entry. Use when preparing a release, generating release notes, or updating CHANGELOG.md.

2 4mo ago A 53 tokens original MIT

checkpoint

39

hamzaPixl/pixl-ai

Skill Claude CodeCodex

Save and restore full working context: branch, uncommitted changes, task state, decisions, and remaining work. Use when switching tasks, ending a session, pausing work, or resuming from where you left off.

2 4mo ago B 47 tokens original MIT

claude-md

40

hamzaPixl/pixl-ai

Skill Claude CodeCodex

Generate or improve CLAUDE.md files. Two modes: (1) Generate — analyze a codebase and create a comprehensive CLAUDE.md from scratch, (2) Improve — audit existing CLAUDE.md files, score quality, and apply targeted updates. Use when asked to create, generate, audit, improve, fix, or maintain CLAUDE.md files.

2 4mo ago A 79 tokens original MIT

hamzaPixl/pixl-ai

Skill Claude CodeCodex

Onboard a new client project: scan the codebase, generate a CLAUDE.md, catalog the tech stack, and create a context packet. Use when starting work on a new project, onboarding a client, or setting up Claude for a new codebase.

2 4mo ago A 58 tokens original MIT

code-reduction

42

hamzaPixl/pixl-ai

Skill Claude CodeCodex

Dead code elimination, dependency cleanup, deduplication, and consolidation. Two modes: (1) Quick scan — runs knip/depcheck/vulture for fast tool-driven analysis, (2) Deep analysis — parallel Explore agents for structural redundancy and gap detection. Use when asked to reduce codebase size, find unused code, clean up…

2 4mo ago A 86 tokens original MIT

code-review

43

hamzaPixl/pixl-ai

Skill Claude CodeCodex

Multi-agent PR review with confidence scoring and AUTO-FIX/ASK classification. Quick mode (3 reviewers) or full mode (8 specialists including red team). Use when reviewing a PR, auditing changes before merge, or running automated review on a branch.

2 4mo ago A 53 tokens original MIT

command-development

44

hamzaPixl/pixl-ai

Skill Claude CodeCodex

Create slash commands for Claude Code plugins. Use when defining reusable prompts with frontmatter, arguments, file references, or interactive workflows.

2 4mo ago A 29 tokens original MIT

content-marketing

45

hamzaPixl/pixl-ai

Skill Claude CodeCodex

Content strategy and hands-on copywriting for business websites: audience analysis, content calendar, SEO-optimized copy, blog plans. Use for strategic content planning or writing individual pieces with human review. For batch-generating content across many pages automatically, use /content-pipeline instead.

2 4mo ago A 59 tokens original MIT

content-pipeline

46

hamzaPixl/pixl-ai

Skill Claude CodeCodex

AUTOMATED multi-step LLM content generation pipeline: discovery → research → generate → translate → validate → assemble. Use when you need to batch-generate content across multiple pages/sections, auto-translate, or run quality validation loops. For content STRATEGY and manual copywriting, use /content-marketing…

2 4mo ago A 69 tokens original MIT

hamzaPixl/pixl-ai

Skill Claude CodeCodex

Template and utilities for creating context packets — the standard format for passing structured data between agents. Use when setting up inter-agent communication or when a skill needs to produce/consume shared context.

2 4mo ago A 42 tokens original MIT

continuous-learning

48

hamzaPixl/pixl-ai

Skill Claude CodeCodex

Instinct-based learning integrated with pixl-crew memory. Observes patterns across sessions, records instincts (heuristics), and applies them to future tasks. Use when asked to learn from mistakes, record a pattern, review instincts, or improve future performance.

2 4mo ago A 54 tokens original MIT