Recently added

One brand system for everything: design tokens, the visual laws, the run-presentation contract and the voice core, generated into every consumer with a CI drift gate.

skillfactory is an open-source Claude Code plugin indexed from GitHub. Review its manifest and source before installing.

shipflow is an open-source Claude Code plugin indexed from GitHub. Review its manifest and source before installing.

resume is an open-source Claude Code plugin indexed from GitHub. Review its manifest and source before installing.

press is an open-source Claude Code plugin indexed from GitHub. Review its manifest and source before installing.

pluginsync is an open-source Claude Code plugin indexed from GitHub. Review its manifest and source before installing.

github-stats is an open-source Claude Code plugin indexed from GitHub. Review its manifest and source before installing.

ghostwriter is an open-source Claude Code plugin indexed from GitHub. Review its manifest and source before installing.

ghostwriter-x is an open-source Claude Code plugin indexed from GitHub. Review its manifest and source before installing.

ghfactory is an open-source Claude Code plugin indexed from GitHub. Review its manifest and source before installing.

eval is an open-source Claude Code plugin indexed from GitHub. Review its manifest and source before installing.

devlog is an open-source Claude Code plugin indexed from GitHub. Review its manifest and source before installing.

city-report is an open-source Claude Code plugin indexed from GitHub. Review its manifest and source before installing.

intact is an open-source Claude Code plugin indexed from GitHub. Review its manifest and source before installing.

Reconcile the plugins installed on this machine with what the marketplace actually offers, and never call a plugin live before the restart that makes it so.

Interactive network security investigation: inspect processes, compare connections, capture scoped packets, and stop user-selected processes with confirmation.

Takes one open GitHub issue to a reviewed pull request with a red-teamed plan, proven implementation, and converging review loop on Claude Code or Codex.

Triages a Gmail inbox against rules you wrote — categorising every thread and trashing only what one of your own rules names, never what the model merely thinks is junk.

Show GitHub profile statistics for a user — total commits, followers/following, stars, pull requests, and issues — as a clean summary table, plus a repository browser and per-repo detail. Use when the user asks for "github stats", "github-stats", a GitHub profile summary, "how many stars/followers/commits does <user> have", repo stats for a GitHub account, or wants to create a repo while looking at stats. Powered by the gh CLI.

Write engaging LinkedIn posts in the user's own voice and publish them to their profile after they approve. Use when the user wants to draft, write, or post something to LinkedIn, asks for a "LinkedIn post", wants content about trending topics in their field, or wants to set up / configure LinkedIn auto-posting. Learns the user's voice from their past posts and never publishes without explicit approval.

Write sharp X (Twitter) posts and threads in the user's own voice and publish them through the free Typefully API after they approve. Use when the user wants to draft, write, or post something to X or Twitter, asks for a "tweet", a "thread", or an "X post", or wants to set up X posting. Enforces X's 280-weighted-character limit per tweet, formats threads natively, and never publishes without explicit approval.

Generate GitHub Actions workflows that are verified rather than hoped for — every action ref resolved to a real pinned SHA, actionlint and zizmor clean, staleness reported, before you ever see the YAML.

Grades a real run of a skill against the contract that skill committed to, keeps only the findings it can cite, and turns the ones that will recur into permanent eval cases.

Turn each new version release (git tag) into a polished, researched how-to guide with real gotchas, written in your voice and published to GitHub

Pull US Census demographic, economic, housing and health data for any American city from the Data USA API, answer questions about it instantly, and generate a polished editorial HTML report. Use when the user asks for a city report, city profile, demographics, census data, population, median income, cost of living, or "tell me about <city>, <state>" — or wants to compare a city against its state and the nation.

Give it just a name; it blind-searches the open web for that person's presence, keeps only what it can prove is them, and renders what it found as a press-styled brand report.

Find the Apple TVs on your network, pair with one, and control it from chat — and never say a command landed until the TV's own state says so.

Incremental verification for Claude Code: rerun affected checks, reuse authoritative passing evidence with revision-aware receipts, and keep an optional Guarded approval boundary for higher-risk work.

Project knowledge graph — captures decisions, failures, discoveries, patterns, and reasoning chains. Fully local, no API keys.

Coordinate tracked work with taskix and read-only Markdown boards.

Connect live Claude Code sessions to Agentix.

Claude-native authorized external recon and ASM methodology bundle with 90 documented capabilities across 11 skills, deterministic Python controls, an optional loopback-only limited-control web plane, explicit human-reviewed finding promotion, and optional fixed policy-gated MCP enrichment. It does not provide unrestricted exploitation or autonomous recon orchestration.

Local FFmpeg video/audio editing for coding agents: 42 tools with a machine-readable contract, capability detection, probe-first / verify-last workflow. No cloud, no API keys.

御社のAIに、電話する人を。初期設定・依頼準備・通話結果の取得と改善をAIから。

Human-grounded subjective labeling in two sibling sides: Label Building calibrates and freezes one identified human's construct into a signed Label Handoff; Label Scanning qualifies executors on sealed human gold, labels the corpus under that frozen handoff, and closes with an independent probability audit.

In-lab human interaction with deployed prediction endpoints — the Claude Code chat is the UI. CONSOLE mode (default): patient-first on-demand inference — list patients, view the full chart, list available models (Endpoint_Set packages + live status), and run 'this model on this patient': the endpoint-predict MCP tool prepares the payload (required tables + trigger record), POSTs to the endpoint, and returns the score verbatim with a data-gaps report. STUDY mode: formal reader protocol (frozen review_bundle.json, blind→assisted session, decision-influence metrics vs gold). Endpoint-agnostic: local Flask, Databricks, SageMaker — same wire contract.

Normalizer utilities for the haipipe pipeline: free text in a cohort's own dialect, resolved against a reference bank into numbers that carry their own provenance. One skill per noun — describe-food resolves a FoodName to USDA nutrition; describe-exercise resolves an ExerciseType to a PA Compendium MET and, when the log stated minutes and the body mass is known, an estimated kcal — and haipipe-norm owns the contract all of them obey.

HTML PPT Studio — author professional static HTML presentations (37 themes (academic-report = house default), 31 layouts, 15 full-deck templates, 27 CSS animations, 20 canvas FX, presenter mode with speaker notes). Includes html-to-svg (per-slide 1280x720 SVG export for PowerPoint import). Vendored from https://github.com/lewislulu/html-ppt-skill.

Skills for the haipipe data + NN + endpoint pipeline. Tier-1 user umbrellas (/haipipe-data, /haipipe-nn, /haipipe-end, /haipipe-project, /haipipe-individual) parse intent and dispatch to per-stage / per-target / per-risk-profile Tier-2 specialists. Covers stages 0-2 (per-individual data contract), 1-4 (project-wide data build), 5 (model training), 6 (endpoint packaging + deploy targets: SageMaker, Databricks, Flask, MLflow), and per-project scaffolding.

Single-entry Obsidian intake and chronicle tools. Use /note to capture readable daily logs, people notes, project/matter notes, tasks, timelines, meetings, emails, and raw material.

JHU Research-IT helpdesk chatbot. Answers how-to questions about Johns Hopkins research IT infrastructure (PMAP, SAFE, SAFER, Discovery, Databricks, Crunchr, Epic Cosmos, REACH) by reading a local WIKI/ knowledge base. Invoke as /jhu-research-it.

Use Codex from Claude Code to review code or delegate tasks.

Writing workflows for developing ideas, drafting, and revising prose while preserving your voice.

Opinionated development workflows and reusable skills for planning, building, reviewing, documenting, and shipping software.

Cost-tiered subagents, haiku-powered git/quick-task commands, design commands that route a request on evidence rather than on its wording and write either a root cause or the trade a system makes on one page, then the decisions that follow routed so that only what needs a person reaches one, then an implementable build spec from verified evidence rather than assumption, skills for squashing a PR branch, surfacing unknowns before implementation, reviewing a plan against the requirements it claims to serve, running a parallel multi-lens review over a branch diff, executing a long change in resumable checkpoints, and building a detail design unit by unit off its own work breakdown with Sonnet implementing and Haiku verifying, a pre-merge comprehension quiz, a design-review-to-verified-implementation pipeline with model tiers per phase, a shared rule set installed by /cai:setup, an append-only ledger that records every stage attempt so a new session can say how many times it has been tried, why it failed and who let it through, an opt-in status line showing the project, branch, model, live /effort level and three remaining-capacity gauges, and a shell-aware safety guard hook.

Jeff is a model-native quality control plane and cooperative workflow protocol, not a security sandbox.

A file-backed, multi-sprint harness for long-running development, with separate Planner, Generator, and Evaluator roles.

Read chess positions by the rules: a move's legality with every reason behind it, draws and their evidence, the named facts, ECO openings, Polyglot books and PGN. No engine.

Board-driven issue workflow: triage, pick the next issue, plan it, implement it in a worktree, adversarially review the diff, open a PR, and drive it to green.

Turn any video into a Chinese-narration recap on ffmpeg + one Xiaomi MiMo API key: understanding (ASR + VLM) -> narration -> cut -> voiceover -> assemble. A bundle of independent skills + a thin orchestrator. Runs on macOS, Linux, and Windows.