Development plugins

Gleam language server for code intelligence, diagnostics, and refactoring.

Create, review, rewrite, and debug repository-specific agents.md, .github/agents/*.md, and *.agent.md files.

Skills for designing, testing, and packaging C# incremental source generators, including pipeline caching, marker attributes, diagnostics, interceptors, and multi-SDK support.

Skills for .NET development, including: C# Source Generator

Project for creating an everything copilot, based on everything claude code but specialized for copilot on vscode

A2A protocol client for Claude Code: Claude hands tasks to non-Claude worker agents over Agent2Agent and picks up their results.

Shared widgets for Claude Code mods: other plugins hand $.kit their state and draw what it returns.

Fortran language server (fortls) for code intelligence and diagnostics

8-bit HUD window above the prompt: LV/EXP, HP = context left, MP = plan usage left, G = session cost, a status row (HP LOW!, REST, SAVED, LEVEL UP!) with rest and save buttons

Tool rows as an 8-bit battle log: a TURN header on each turn's first row, HIT, CRIT with +added −removed lines, MISS with the lines that say what failed, COMBO ×n and BREAK, each guard's own verdict (TRAP!, BARRIER!, LOOP!), and SUMMON on the run-in-background pill

Stops tool calls that would leak a secret: literal keys in commands or source files, cat .env / echo $API_KEY, reading key files into the conversation

A deterministic command and file-change gate for coding agents.

InDevYoo

Lightweight spec-driven development: raw task, spec, review gate, plan, code review and delivery manifest, saved as files in your repo.

Check whether a website is ready for AI agents, and fix what fails: 23 checks, a level from 0 to 5 and a guide for each failure. Runs on your machine and can scan localhost. Installs its own binary on first use.

Reverse engineer Android apps with agents, from APK structure down to native libraries.

Install the unified collection of 2565 distinct skills, all authored, covering all 31 profile languages and specialist engineering workflows.

200 skills

Take a change from idea to production, in one plugin. Decide: interview-me, grilling, spec, requirement-convergence. Shape: ddd runs the nine-step DDD Starter Modelling Process with validators, a review page, diagrams and recorded decisions; Blueprint renders architecture, sequence, workflow, data-flow and lifecycle diagrams; ADRs, contracts, module design. Understand: core primes a repo with a knowledge graph and keys keeps one AGENTS.md tree; trace, research-codebase, deepwiki, explain-diff. Build: a planned pipeline from brainstorm to change-description, with tdd, debug, swarm and unlazy. Verify: code-sweep review, complexity, blast radius, verify-feature, harnesses and test suites. Secure: security-sweep. Ship: ship-gate, SLOs, flags, CI/CD, migrations. Principles and stack references are router skills that load one guide at a time. The DDD and Blueprint tools require Node.js 18 or newer. Vendored skills are credited in NOTICE.md.

142 skills · 5 agents

Full-SDLC automation for React/TypeScript frontend engineering: GitHub issue creation, BMAD-method planning, bmalph/Ralph autonomous implementation with Claude CLI, multi-skill code review with BMAD FR/NFR gate plus a mandatory accessibility gate, visual/E2E/Lighthouse QA verification, CI auto-fix, and AI review comment resolution — looping until the PR is fully finished.

86 skills · 10 commands · 7 agents

Moonweave AI governance, planning, RFC, engineering, security, quality, release, research, and knowledge management skills.

27 skills · 75 commands

The Claude Code stack, guided: init installs it from scratch into a project and, typed again after the restart, bootstraps it once (tools, status line, captures, AGENTS.md), update refreshes an existing install and prunes what upstream removed, configure adjusts the selection (add or drop), validate reconciles an install to the project - prunes what its framework does not use and adds what the detected stacks are missing - all drive a dependency-complete selection with a prerequisite check; status shows the install read-only, per-area tables; uninstall removes what the install ledger says the stack wrote, and nothing of the user's.

101 skills

Agentic OS for Claude Code. 39 commands, 42 skills, 10 OS capabilities. /forge dispatches specialized agents across a 7-phase build pipeline. /dfe runs 6-pass adversarial code review (5 domain passes + 1 adversarial pass). /aisle reports security posture and routes scans while the old fail-closed gate remains shelved. /recall searches connected cross-session memory. /hud renders live OS state in the statusline. One install, one config.

42 skills · 39 commands · 17 agents · hooks

Vulpora 에이전트 29종과 스킬 66종을 설치하는 catalog 플러그인.

66 skills · 29 agents

A developer workflow for coding agents: skills for every stage, review subagents, hooks and repository templates. Required.

85 skills · 8 agents · hooks

Seven release-adjacent skills selected from Derek Pearson's 86-skill iOS collection, chosen because nothing else on this list covers them: App Clips (own store record and size limit), crash symbolication, DeviceCheck/App Attest, String Catalogs and localized metadata, simulator driving for capture, MetricKit hang and crash diagnostics, and StoreKit 2 IAP. NOT MIT — PolyForm Perimeter 1.0.0, source-available with a noncompete clause. Read docs/companion-skills.md before installing. Third-party skills by Derek Pearson.

86 skills · 1 MCP server

Coding standards, review workflows, and an issue-to-PR agent roster for PHP and Laravel projects

67 skills · 4 commands · 9 agents

The principles-driven stack materializer — implements the vwf stack-adapter contract with a dispatch rule that runs per component: a component a shipped pack covers is copied verbatim; an uncovered one is generated (per-topic Context7 research → vwf's principles catalog → the stackgen-skill-reviewer gate, capped at four rounds). A materialization lands in the repo's committed .claude/ tree — skills, agents, hooks and rules — plus, for packs that ship one, a config/ tree mirroring the repo root; language servers go to the generated local plugin vwf-tools outside the repo, together with every MCP server a component declares, never into a repo file. Every artifact is shaped by a closed kind vocabulary, recorded per component in a lockfile, editable and plugin-independent for collaborators. Re-sync is explicit, per component and lockfile-diffed, never silent. Ships curated packs and bundles across a closed kind vocabulary, inventoried in stacks/inventory.md; a bundle is a recorded composition of component refs, never a directory. Three tool axes (design, cicd, stylesheet) let a pack be reached by the config token that selects it, which is what rescued a CI pack no menu could offer; optional tools (context7, notion, taste-skill) are an addon axis the user picks. assets/artifact-doctrine.md carries the host rules deciding whether a generated skill, agent or hook is valid at all, and assets/contracts/ the provider-neutral capability doctrine each category is compared against. .claude/settings.json is written only behind its own explicit consent line, merging never owning.

78 skills · 1 agent

Engineering craft toolkit: TDD, testing, refactoring, architecture, delivery and writing skills, agents and commands.

66 skills · 7 agents · hooks

BYDFi-oriented Claude Code workflow for Next.js, React, TSX, TypeScript, reviews, audits, build fixes, and PR workflows.

26 skills · 33 commands · 5 agents

Daniel Bodnar's personal skill set (28 skills).

28 skills · 35 MCP servers