Development plugins

The Viglet Design System in an agent session: the page contract as a skill, a check command over the package's gates, and a hook that catches a local copy of a shared component as it is written.

1 skill · 1 command · 1 MCP server · hooks

Teaches the LLM how best to maintain a fork.

4 skills

Leader-only orchestration plugin with hooks, skills, bridge dispatch, GPT-Pro background tasks, and E2E evidence gates.

3 skills · hooks

Continuous self-referential AI loops for iterative development (Ralph Wiggum technique). Supports concurrent sessions with --name option.

3 commands · hooks

Helps non-developers vibe code without bugs; design, clarify, and then develop features.

2 commands · 2 agents

Web development track: Next.js, Supabase, Vercel, Vercel AI SDK, Gemini, Polar, Resend

4 skills

Multi-agent collaborative sandbox. Spin up AI agent teams that chat together in a shared Matrix room — each with their own identity, memory, and tools.

4 skills

Autonomous experiment loop — try ideas, measure results, keep what works, discard what doesn't, repeat forever.

1 skill · 2 commands · hooks

Godot 4 guardrails for Claude Code: Godot-3-ism linter (92 sourced rules), scene map with real node paths, version pin into CLAUDE.md. Verified on Godot 4.7.2.

3 skills · hooks

Spec-Driven Development workflow with ticks-based task tracking for AI agents

4 commands

Backend engineering assistant with GitHub, Jira, and Context7 (Laravel) MCP integrations

1 command · 3 MCP servers

Long-running autonomous development loop with structured progress tracking. Combines Ralph Loop iteration with Anthropic's initializer + coding agent pattern.

3 commands · hooks

Specification gate between Task Master expansion and an existing TDD workflow: routes each task to PASS/GHERKIN/DMN/BOTH, reviews it in an independent context, and emits a READY-only TDD handoff. Does not modify nestjs/nextjs-dev-workflow.

2 skills · 1 command · 1 agent

JS→TS migration syntax-equivalence checker. Strips TS-only constructs and structurally compares the result to the JS baseline on the default branch. Surfaces findings to the agent so it self-corrects mid-migration.

1 skill · 1 command · 1 MCP server · hooks

Persistent knowledge base and other skills for coding agents in long-running dev sessions

4 skills

Create, deploy, and manage Power Automate cloud flows as code (JSON definitions)

3 skills · hooks

Hokusai task routing for Claude Code - get an Anthropic model recommendation before you spend inference budget.

4 commands

Developer skills: /tldr and /tldr-extended collapse a long or technical Claude Code session into a plain-English summary with one committed recommendation; /fable-review is the fresh-eyes final review before work ships, with reviewers pinned to the bundled fable-reviewer agent (Opus, effort high).

3 skills · 1 agent

Cross-model consultation, plan review, and code review. By default, Claude Code uses Codex and Codex uses Claude Code. The Cursor Agent CLI is available as an optional reviewer.

3 skills · hooks

C#/.NET documentation skills (Russian output): doc-cs (inline XML-doc), doc-md (external Markdown, one file per class), doc-db (Oracle schema reverse-engineering from ADO.NET code), doc-folder (single in-place README.md folder overview).

4 skills

Build reliable, sandbox-first integrations with Syntage

4 skills

implement-change, clean-architecture, write-tests and create-readme.

4 skills

Personal development toolkit with Tavily web search MCP, deep-research command, explain-code skill, and better-search hooks

1 skill · 1 command · 1 MCP server · hooks

Test-driven development workflow with red/green/refactor phases and coverage monitoring

3 skills · 1 command

Search a shared base of AI-research experiences (what failed, what worked, and why) before designing experiments, choosing configs, debugging runs, or driving an autoresearch loop — and record new experiences afterward.

1 skill · 3 commands

7-phase pipeline for AI context engineering: define the problem → collect candidates → select → structure → compress → generate output → verify

4 skills

Deep codebase research and impact analysis before making changes. Iteratively explores code, maps all dependencies, identifies risks and critical points, and creates a comprehensive implementation plan.

3 commands · hooks

Shared Switchboard gateway, setup, status and reversible Claude startup integration.

4 skills

Personal coding guardrails: fundamental dev principles (DRY/YAGNI/KISS/SINE) + warn-only worktree-gate hook.

2 skills · 1 command · hooks

Replaces raw Edit / Write / MultiEdit with twenty-one kind-specific edit_* MCP tools (15 SQLite-derived + edit_cosmetic + 5 workflow-axis kinds) that encode the testing obligations for each kind of change directly in their tool descriptions. Every declaration carries a required provenance field. Impl tools carry a required prod/test target flag.

2 skills · 1 MCP server · hooks