memory-cognition-provider

Use when planning or implementing a new app, feature, automation, agent, integration, or roadmap where substantial work could accumulate before any useful…

Use when creating, redesigning, or choosing tooling for polished PDF documents, especially editorial, educational, academic, proposal, report, workbook…

Use when the user sends /persist or an explicitly authorized AI conversation must supervise a long-running workflow across chat context limits, session…

Use when designing, reviewing, or implementing LLM-powered applications, agents, RAG systems, multimodal apps, generative UI, voice agents, always-on agents…

Use when a task is large, long-running, repository-heavy, multi-agent, reference-heavy, or likely to waste tokens by loading, rereading, rediscovering, or…

Use when work spans multiple steps, sessions, interruptions, branches, agents, or long-running implementation where execution facts must survive context loss.

Use when creating, reviewing, or refining visual assets such as hero images, galleries, portfolio thumbnails, or editorial imagery.

Use when an authorized Windows PC must be observed or controlled through its interactive desktop, including screenshots, windows, visible UI, mouse, keyboard…

Use when the user sends /revise, starts a message with /revise, or asks for a rigorous repository/code review with verified findings.

Use when making codebase changes that could affect multiple files, state, or existing approved work.

Use when documentation, specs, APIs, or external references influence implementation decisions.

Use when wrapping up a work session, pausing a project, or preparing a clean resume point for future sessions.