Recently added

macOS sleep prevention toggle with battery and network monitoring

Preflight the local environment before agent-device attaches to a physical iOS or Android device: device selector resolution, lock-state check, adb authorization, signing team derivation, device registration, DDI mount, DevToolsSecurity, helper install, runner cache and daemon env staleness

Engine aware outbound writing: LinkedIn posts in the correct brand voice per engine, and the newsletter writer that interviews once then drafts every edition without questions. Both draft only. Neither publishes.

Throwaway test plugin. Its single skill prints one known phrase, to prove the marketplace round trip works in both Claude Code and Cowork.

Cross engine operating skills: the daily focus check-in, clean session handoff, epistemic honesty mode, the greenlight council for pressure testing a decision, and the OS operator and optimizer. Note that os-operator, os-optimizer and skill-adapter contain hardcoded Windows paths and are Claude Code only until reworked.

Global Exporters operating team: the email command centre, freight quote intake into Airtable, and the hard gated verification protocol for the Command Centre app.

The Gareth Cohen Experience operating team: bookings, cold outreach, command secretary, finance controller, risk reviewer, SEO and website, social campaigns, the daily agency gig sweep, and the annual sole trader tax return.

Project memory for Claude Code and Codex. Storybloq stores tickets, issues, handovers, and lessons in a .story/ folder inside your repo.

UiPath Maestro preview skills — the builder-SDK authoring path for uipath-maestro-{flow,case,bpmn}. Authored here; loads as a Claude Code plugin root (skills/<name>/SKILL.md).

UiPath skills for building, validating, deploying, and operating automations and agents from Codex.

Your lore, across time and space. Recall and learn with your private signet.

律师工作台 3.0：破产业务先行的法律工作台，两个入口加一个路由，四个参考 skill 由模型自行够到。

Build, inspect, and debug UI components using @barefootjs/cli (>= 0.11.0). Covers component discovery, scaffolding, signal inspection, reactive performance profiling, and previews.

Task-specific expert council and real-world decision support that preserves named human authority.

Task and activity orchestration for AI coding agents. Adds Orbit's MCP task and knowledge surface plus workflow skills to Codex.

An engineering workflow for Claude Code. One idea goes in; a reviewed, integrated commit comes out.

Synthesize project, calendar, CRM, and Cortex context into concise weekly client-status drafts for user review.

Find recent stories, rank them for a configured audience, and assemble a voice-matched roundup with traceable sources.

Create an engagement plan, portable AI workspace prompt, folder blueprint, next action, and optional Cortex nodes from approved inputs.

Review relationship context, cooling periods, and positive moments to prioritize referral actions and draft voice-matched asks.

Turn calendar events into a local, reviewable time log and generate invoice drafts from configured client billing rules.

YAGNI enforcement at the harness level for real engineering work — nudges toward the simplest solution and gates against unnecessary dependencies, file sprawl, and unchecked build growth.

YAGNI enforcement at the harness level for real engineering work — nudges toward the simplest solution and gates against unnecessary dependencies, file sprawl, and unchecked build growth.

YAGNI enforcement at the harness level for real engineering work — nudges toward the simplest solution and gates against unnecessary dependencies, file sprawl, and unchecked build growth.

YAGNI enforcement at the harness level for real engineering work — nudges toward the simplest solution and gates against unnecessary dependencies, file sprawl, and unchecked build growth.

YAGNI enforcement at the harness level for real engineering work — nudges toward the simplest solution and gates against unnecessary dependencies, file sprawl, and unchecked build growth.

YAGNI enforcement at the harness level for real engineering work — nudges toward the simplest solution and gates against unnecessary dependencies, file sprawl, and unchecked build growth.

Run real-browser end-to-end (E2E) acceptance tests on your web app from any AI coding agent. Generate test scripts from plain English, replay them on localhost, capture screenshots, and validate user flows like signup, checkout, and dashboards. Works across Claude Code, Cursor, Codex, and Windsurf.

開發流程慣例，除發版流程（.NET / NuGet 專屬）外與程式語言 / 框架無關：計畫文件（預設 docs/plans/，可依 repo 約定改放 gitignored 目錄）的撰寫與格式、定案後交接給接手的 session（不限程式碼實作，交接文件也不限是 plan）、實作期間的驗證閘門（plan 版本確認、範圍對帳、平行路徑檢查）、Claude Code 設定檔語料的定期健檢（含 plugin 版本漂移與自有 marketplace 的清單一致性）。CHANGELOG 草稿整理（依既有慣例偵測後沿用）。.NET / NuGet 的發版流程（前置檢查、版號判定、出貨基準合併 —— PublicAPI 與 analyzer release tracking 兩份、分兩次推送）。後續擴及 CI 與源碼掃描。

Replaces the spinner with an animated ASCII slot machine: pull the lever on every prompt, watch the reels spin while Claude works, cash in when the turn completes.

Org-mode task tracking for Claude Code: org_* MCP tools via a Doom Emacs module, a queued state/clock event architecture reviewed by a human, session-tracking hooks, and the org skill with its conventions as references.

Command-free, skills + subagents plugin for fast single-feature development: digest a plan, get approval, isolate in a worktree, implement, and open a PR — Sonnet build / Opus review, with graded effort.

Higher Agents for Codex: rigorously plan, implement, review, and merge one feature with controlled subagents and fail-closed gates.

Ask a fresh Claude session to review branch and local changes, implement a milestone plan test-first in an isolated worktree, and provide the bounded internal second opinion used by ca dual review.

Claude Code Higher Agents for building ONE feature thoroughly: deep red-teamed plan to human approval, per-task implementation review, a risk-scaled pre-PR gate, independent PR review, rigorous feedback handling, gated merge, conflict resolution, and safe worktree cleanup. Requires the superpowers plugin; model-agnostic.

Ask a fresh Claude session to review branch and local changes, implement a milestone plan test-first in an isolated worktree, and provide the bounded internal second opinion used by ca dual review.

Ask a fresh Codex session to review branch and local changes, or run the ca feature loop: plan with Codex, review milestone and final PRs, synthesize findings, and provide gated merge, conflict resolution, and safe worktree cleanup.

Capture and enrich innovations from conversations or technical sources, request patent preparation, and respond to Lightbringer report and patent-draft reviews.

Carries a pull request from opened to landed: draft state that tracks convergence, same-cycle review-comment triage, line accounting against the stated purpose, and a human hand on the merge by default (an owner-applied merge-when-green label is the standing approval).

Repository documentation that skims well and names real code: the device picked from the content, a paragraph-cap and path probe, fact-preserving restructures, and the architecture page whose boxes name real symbols.

Keeps the Codex desktop Chrome plugin driving Brave and other Chromium browsers: cache sync after app updates, browser-only auth under a custom model provider, and browser-family selection.

Use Wannalead's LinkedIn workflows through its hosted MCP server.

Route every Claude Code tool call through a nucleus Firecracker microVM under an information-flow permission lattice, with a signed mediation receipt per call.

Контроль качества разработки 1С:Предприятие (BSL): стандарты #stdNNN, принципы структурного и объектно-ориентированного программирования, паттерны проектирования применительно к 1С, корректность XML-структур метаданных. Блокирующие гейты не дают объявить работу законченной, пока проверки не прогнаны, а машиночитаемый след фиксирует, что именно проверено и что пропущено.

Token-efficient WB CLI operations and safe multi-repository workflows

Generic GitHub workflow commands for issue management, PR creation, code review, and releases. Works with any repository by auto-detecting settings.

Skill-driven workflow plugin for GitHub development with excellence-by-default quality gates. Encodes team knowledge as composable skills, enforces safety through hooks, and compounds learning across sessions. Strict TDD, Stranger Test, holdout validation, and evidence-based verification built in. Durable goals, workflows, triggers, and runs at .flow/. Requires bash, git, gh, jq, and python3 with PyYAML; on Windows, Git Bash (see docs/windows-support.md).

Evidence-first project documentation system. Produces an audit-ready 23-file package for technical due diligence, engineering and product onboarding, and partner or customer communication — every material claim traced to a ledger row and a claim state, three context-isolated verification passes, and a binary release gate. Ships a post-merge GitHub Actions job that keeps the package current by opening a documentation PR. Requires bash, git and jq; on Windows, Git Bash (see docs/windows-support.md).

NCI (Narrative Credibility Index) manipulation detection with integrated deep research fact-checking. Analyzes content for propaganda, disinformation, and manipulation patterns across 20 categories. Includes claim verification using deep research methodology.

Structured product development methodology with traceable evidence, explicit decisions, and constrained spec generation. Forces auditability through semantic IDs, quality gates, and impact-aware updates.