Connected-Mate

Generates a secure, branded, organization-specific launcher that wraps Claude Code, Codex CLI, Gemini CLI, Cursor, or Cline onto a corporate AI gateway, then helps the user distribute it to their team. Trigger phrases include "corporate launcher", "wrap…

Generates a secure, branded, organization-specific launcher that wraps Claude Code, Codex CLI, Gemini CLI, Cursor, or Cline onto a corporate AI gateway, then helps the user distribute it to their team. Triggers when a user says their company "does not…

Generates a secure, branded, organization-specific launcher that wraps Claude Code, Codex CLI, Gemini CLI, Cursor, or Cline onto a corporate AI gateway. Triggers on "wrap claude for my company", "white-label cline", "internal AI CLI", "bedrock gateway",…

Reasoning discipline for hard, multi-step tasks — how to decompose ambiguous problems, verify work before reporting it done, and choose the next action. Use at the start of any task expected to take more than a few steps, when a plan is failing, when…

Design and run autonomous Claude Code loops safely — the 5-component pattern (Discovery, Handoff, Verification, Persistence, Scheduling), exit gates, budget caps, and guardrails. Use when setting up /loop, /schedule routines, overnight agents, ralph loops, or…

Multi-agent orchestration discipline — when to spawn subagents vs agent teams vs staying solo, how to write bounded task prompts, effort budgets by complexity, and coordination anti-patterns. Use before fanning work out to agents for research, implementation,…

Install and configure AnnotateKit in an iOS or SwiftUI application so on-device visual annotations are sent directly to Codex through the bundled MCP bridge. Use when adding AnnotateKit, enabling iOS UI annotation, or setting up the AnnotateKit workflow.

Read and implement pending AnnotateKit iOS or SwiftUI visual feedback delivered by the AnnotateKit MCP connector, then resolve each accepted annotation.

Preview an isolated, privacy-preserving route for a Codex coding task. Use when a user wants a recommendation between economy, balanced, and quality tiers before sending a first prompt.