Automation plugins

Mobile auto-shipper. Drive a chat request through analyze -> implement -> QA -> PR on a real simulator/emulator/device. The orchestrating agent is the coder — no separate engine CLI. Includes backend/Supabase and Flutter QA setup.

6 skills

Save state + auto-restart helper for atomic /loop sessions.

2 skills · 2 commands · hooks

Autonomous maintenance toolkit: drift-check runner against context-file verification commands, morning-briefing template, Windows Task Scheduler registration helper, auto-memory housekeeping for ~/.claude/projects/*/memory/.

2 skills · 1 command · 1 agent · hooks

Event-driven routines for GitHub: implementer (issues.opened → focused PR; gate-conditional landing — squash auto-merge only where the target branch has required status checks, else the PR stays open) and optional merger (pull_request.opened → six-condition fail-closed gate → gh pr merge --auto or one needs-you escalation). Author + label gated, repo-agnostic skills.

4 skills

Arms a watchdog that powers the machine off once every Claude session on it has gone idle.

3 skills · hooks

Interactive crawler development workflow using Chrome DevTools MCP. Explore sites with user, document paths, build and test crawlers.

1 skill · 1 command · 2 agents

Run a prompt/command on a recurring loop for long-running automation.

3 commands · hooks

夜间循环：夜间计划与晨间固化评审。Night loops: nightly planning and morning crystallization review.

2 skills · 2 commands

Connects Claude Code to the Synthflow AI voice-agent platform through its hosted MCP server, with skills for reviewing calls and auditing agent prompts, plus a docs-search connector.

2 skills · 2 MCP servers

Fenced Brain/Hand coding loop with read-only planner, sandboxed executor, and destructive-command guard hooks.

2 agents · hooks

Direct CDP browser control for agents: coordinate clicks, screenshots, persistent Python session, local Chrome or Browser Use cloud. Ships skills only; the `browser-harness` CLI is a one-time install prerequisite.

3 skills

Appmixer connector development skills: scaffold/build connectors, CLI component tests, and component review against Appmixer standards. Pure instructions — the appmixer CLI is the only external tool.

3 skills

Handle Telegram messages for a personal bot. Check for new messages, process requests, and send responses back via the Telegram Bot API. Supports authorized single-user mode with automatic message filtering and error handling.

3 skills

Real Chrome browser automation via MCP relay. Anti-bot bypass, login session reuse, real fingerprints. Multi-user with per-user auth tokens.

1 skill · 1 agent · 1 MCP server

Rules for multi-agent work: model routing per phase, fan-out sizing, closed output schemas and citation checks, with CLI workers for search and extract and hooks that load the rules and check routing at dispatch.

1 skill · hooks

Run commands, builds, and tests headlessly on a remote dev box over SSH, with per-user host configuration.

1 skill · 1 command

Web browser automation for Claude Code and AI agents. Headless or headed mode, persistent auth, stealth mode, video recording.

1 skill · 1 command

Another vendor's agent CLI as an isolated subagent, from any host: ask a read-only question or hold a multi-turn conversation (agy, opencode, Codex, Grok, Hermes, Copilot, Cursor, Muse, Kimi, or Claude Code from another host), get a second opinion on a diff, or delegate one coding slice to agy, opencode, Copilot, Codex, Grok, Cursor or Claude Code under the worker CLI's own hooks and a mechanical gate. Tiers (cheap, mid, top) map to a model and effort per backend. The verdict never reads the CLI's exit 0 as success.

1 skill · hooks

A software factory on your own box: Trello cards labelled `ready` become one PR per step, at most three in flight. A dispatcher on a timer runs one background Claude worker per step; you approve each spec and review each PR from the Claude app, pinged on Slack. Only the tiniest PRs auto-merge, through a deterministic gate hook.

1 skill · hooks

Browser automation via Puppeteer - execute JavaScript code with direct access to browser instance

1 MCP server

Discover, monitor and safely switch TP-Link Kasa and Tapo smart plugs. Requires the powerctl command; see the repository README for installation.

1 skill

Control separate Claude Code sessions from this chat via playwright-cli

1 skill

Run a multi-agent feedback loop within this session using the built-in Agent tool

1 skill

Connect an existing Windows-MCP instance (RDP + Windows MCP) to the Strawberry terminal sidebar and tasks

1 MCP server

Super Marius — let an agent click, type, and screenshot on macOS. Hammerspoon-based desktop automation, Tahoe-compatible.

1 skill

Orchestrator skill — never does work itself, always dispatches Opus subagents (architect → tester → developer → reviewer). Two modes: interactive (at-keyboard) and away (overnight cron triage with risk gates).

1 skill

Continuous PR monitoring with auto-review, auto-fix of review comments, and label management across repositories

1 skill

Telegram bot bridging to Claude Code CLI — persistent sessions, images, /back, /stop, /yes /no confirmation, full permissions

1 skill

Subagent driving an approved change from branch → PR → external review → merge. External reviewer: GLM 5.3 via a read-only opencode `review` agent (repo access); falls back to the `pr-review` subagent pinned to GLM 5.3, then /code-review. Triggered only after the user has approved a plan and authorized full automation through merge.

1 agent

Subagent for repo documentation cleanup and AI-agent instruction consolidation (AGENTS.md / CLAUDE.md / GEMINI.md / Copilot). Nine-phase workflow that scales to the task. Removes stale artifacts without changing application behavior. Output is a PR, not a merge.

1 agent