发现 Skills

Automate Electron desktop apps (VS Code, Slack, Discord, Figma, Notion, Spotify, etc.) by connecting agent-browser to the app's Chrome DevTools Protocol (CDP) port. Use this when you need to interact with an Electron app, automate a desktop application, connect to a running app, control a native app, or test an Electron application. Typical triggers include "automate Slack app", "control VS Code", "interact with Discord app", "test this Electron app", "connect to desktop app", or any task that requires automating a native Electron application.

This skill should be used when the user asks to "analyze competitors", "assess competitive landscape", "identify differentiation", "evaluate market positioning", "apply Porter's Five Forces", or requests competitive strategy analysis.

Design durable workflows with Temporal for distributed systems. Covers workflow vs activity separation, saga patterns, state management, and determinism constraints. Use when building long-running processes, distributed transactions, or microservice orchestration.

Use this skill when implementing tasks according to Conductor's TDD workflow, handling phase checkpoints, managing git commits for tasks, or understanding the verification protocol.

Multi-platform content distribution across X, LinkedIn, Threads, and Bluesky. Uses a content-engine pattern to adapt content for each platform. Never posts identical content across platforms. Intended for use when a user wants to distribute content across social platforms.

An evidence-first audit workflow for ECC Tools burn and billing. Used to investigate runaway PR creation in the ECC-Tools repository, quota bypass, paid-model leakage, duplicate jobs, or GitHub App cost spikes.

GitHub operations, automation, API integration, and CI/CD workflows.

Japanese translation: This file requires a Japanese translation for lead-intelligence.

Build a reusable Manim interpreter for technical concepts, charts, system diagrams, and product demos, intended to be handed off to a broader ECC video stack when needed. Use when the user wants a clear animated explanation rather than a generic talking-head script.

Japanese translation: This file requires a Japanese translation for motion-advanced.

Diagnose interface errors, drops, CRC errors, duplex mismatches, flapping, speed-negotiation problems, and counter trends on routers, switches, and Linux hosts.

Security best practices for Quarkus authentication, authorization, JWT/OIDC, RBAC, input validation, CSRF, secret management, and dependency security.

Cross-stack source-code asset audit — classifies each file, detects embedded third-party libraries, and provides a practical four-tier assessment for each module together with an interactive HTML report.

Accessible concurrency in Swift 6.2 — by default execution remains single-threaded on the caller's actor; use @concurrent for explicit background offload, and use MainActor types to preserve isolation consistency.

name: Token Budget Advisor (token-budget-advisor) description: Before answering, present users with informed choices about how much of the model's response budget (depth/length) to consume. Use this skill when the user explicitly wants to control response length, depth, or token budget. Trigger conditions: "token budget", "token count", "token usage", "token limit", "response length", "answer depth", etc. Behavior: estimate input tokens heuristically (prose: words × 1.3; code: chars ÷ 4), apply a multiplier range by prompt complexity (simple / medium / medium-high / complex / creative) to calculate the available response window, and present four depth options (25% / 50% / 75% / 100%) with approximate token counts and expected output characteristics. Use cases: save API costs, request shorter or more detailed answers, control code snippet output, avoid truncation or overly long replies. Will not trigger when a session-level setting already defines depth or when the request clearly expects a one-word answer. Benefits: improved cost predictability, greater user control, and better response quality (heuristic accuracy ~85–90%).

End-to-end (E2E) testing for Windows native desktop applications (WPF, WinForms, Win32/MFC, Qt) using pywinauto and Windows UI Automation.

An evidence-based automation inventory and overlap-audit workflow for ECC. Use this when users want to know which jobs, hooks, connectors, MCP servers, or wrappers are live, broken, redundant, or missing before remediation.

Patterns and architectures for autonomous Claude Code loops — from simple sequential pipelines to RFC-driven multi-agent DAG systems.

Bun as a runtime, package manager, bundler, and test runner. Guidance on choosing Bun vs Node, migration notes, and Vercel support.

Tracks user-facing buttons and touchpoints through the full sequence of state changes to find bugs where handlers work individually but cancel each other out, produce an incorrect final state, or leave the UI in an inconsistent state. Use cases: when systematic debugging has found a bug but users report a broken button, or after a major refactor that touches the shared state store.

Playwright E2E testing patterns, Page Object Model, configuration, CI/CD integration, artifact management, and flaky-test strategies.

An evidence-based mailbox workflow for ECC: triage, draft creation, send verification, and safe follow-up of sent mail. Use this when users need to organize mail, prepare drafts or send through real mail clients, or to prove what arrived in sent mail.

Japanese translation: This file requires a Japanese localization for mcp-server-patterns.

Japanese translation: this file requires a Japanese translation for motion-foundations