Productivity plugins

Model/effort tiering for every delegated agent: tier-ladder skill, deterministic PreToolUse gate, per-delegation ledger, live canary.

1 skill · 1 command · hooks

Turn study materials into a searchable AI notebook with /ingest, /ask, and /view.

3 commands

Measures how much you actually steer your AI agent versus letting it drive, and offers a daily report.

2 commands · hooks

Three skills for working with Claude Code: context-budgeted subagent delegation (fanout), driving your own logged-in Chrome (my-chrome), and running a task to completion without check-in pauses (autopilot).

3 skills

Deterministic discipline gates for agent output and edits, with the adw-cli and agent-discipline terminal tools included.

2 skills · hooks

Build new ad beats for a Riverio campaign: clone a canvas, run it on Riverio (quoted first), edit in Palmier Pro, QA and upload the finals.

1 skill · 2 MCP servers

Automatically loads your personal context doc from Google Drive at the start of every session.

2 skills · hooks

Feeds your Wybe node from Claude Code: a background distil turns your sessions into short work facts, redacted on your laptop before anything leaves it. The node never reads your machine. Sends only, and never reads what your colleague knows; two things come back: a task you gave your colleague for an open session (one that needs your yes is shown to you first), and your colleague's open questions, shown at session start for you to answer or not.

1 skill · 1 MCP server · hooks

Build, publish, and operate Dashform AI forms from Claude Code. Connects the Dashform MCP server (getaiform.com) and adds form-building workflows.

1 skill · 1 command · 1 MCP server

Appends a short, plain-language summary under every long reply: the most important points first, as a list, for a reader with no time.

2 skills · hooks

Collaborate in Flow threads and use Voice Mode from Claude Cowork and Claude Code.

2 skills · 1 MCP server

Pre-registers experiments, runs them unattended while you sleep, adjudicates results with a blind subagent, and leaves a morning brief. No code, no daemons, no MCP — three markdown files.

1 skill · 2 agents

Configures Claude Code from the hstack repo and adds review-until-convergence and parallel worktree agents.

2 skills · 1 command

外部创意驱动的中文单元写作与证据化章节分析插件

3 skills

Analyze a GitHub repo and generate a complete launch kit — each as its own file: analysis, release notes, changelog, blog, and X/Reddit/Hacker News/LinkedIn posts, plus an SVG share card

2 skills · 1 MCP server

Project planning skills: notes capture, plan navigation, sprint planning, bootstrap, docs-update, and PR review.

3 skills

Reads your recent Claude Code transcripts, keeps only the turns you typed yourself, and gives a new session the threads still open. One line at session start; the full read on /ketchup.

1 skill · 1 command · hooks

MCP server + commands for Plaud voice recording pipeline: sync, classify, status, health, transcript, quota.

1 skill · 1 command · 1 MCP server

Visuelles Status-Signal: Terminal-Hintergrund leuchtet auf, sobald Claude Code auf Input wartet, und resetet automatisch sobald wieder Tools ausgeführt werden. Farbe via /terminal-color frei wählbar.

2 commands · hooks

Structured problem-solving with an explicit solution tree and real-time visualisation.

1 command · 1 MCP server · hooks

Session continuity — hand off context between sessions to split tasks, coordinate work, or resume later.

3 commands

Records subagent lifecycle events so subherd can show exact start and stop times and link each session to its herdr pane

2 commands · hooks

Give each Claude Code profile its own startup mascot — tie, pirate hat, crown, halo, shades, or a smaller companion holding its hand — so several terminals are told apart at a glance. Re-applies itself whenever Claude Code updates.

1 skill · 1 command · hooks

Unique per-session agent names + iMessage routing for Claude Code and Hermes. Every session picks its own callsign at start.

1 skill · 1 command · hooks

Write plain, human-sounding technical and professional prose: merges SimpleEnglish's ASD-STE100 discipline with agent-style, humanizer, and sepia's AI-tell catalogs into one non-fiction ruleset.

2 skills · hooks

Persistent memory layer for AI agents. Store, recall, search, and share memories with semantic search. Triggers on 'remember', 'recall', 'search memories', 'update memory', 'share', 'subscribe', 'permissions'.

1 skill · 1 agent · hooks

Documentation Engineer skill: judges what is worth recording, scores its value, and — with your approval — writes Korean engineering docs (Troubleshooting, ADR, TIL, Retrospective, Resume Material, Study Note) into your knowledge vault.

1 skill · 1 command · hooks

Personal development hooks, helper scripts, workflow caveats, and code-quality guidance.

3 skills

Every Claude Code session joins the miragen memory bridge: working state and selective recall injected at start, prompts and outcomes captured, episodes filed as Loimi artifacts, plus memory_* and store_* MCP tools. Works locally and in cloud sessions.

1 skill · 1 MCP server · hooks

Identity, tasks, policy, memory, and provenance for AI agents.

3 skills