Skip to main content
Before sending a prompt, select a mode from the dropdown in the chat input bar. Each mode tunes the Manager’s behavior and the style of every agent’s output for that task. The same request can produce very different results depending on which mode you pick.

​
Available modes

General assistant for development and knowledge tasks.The default mode. Agents respond with precision and technical depth, targeting approximately 400 words for standard requests and scaling up only for complex architectural tasks. Output defaults to structured markdown with actionable code blocks.Use Standard for everyday tasks: drafting a note, answering a question, making a quick edit, or any task where you want a direct, well-structured response without a specific methodology applied.
Pedagogy-first: clarity, rigor, and careful truthfulness.Agents apply a scaffolded learning approach to every response — starting with intuition and concrete analogies before introducing formal definitions. Uncertain claims are flagged explicitly with [needs verification] rather than stated as fact.Every response includes a ## Key Takeaways section with 3–5 bullet points summarizing the most important ideas. Common misconceptions about the topic are addressed proactively. Notes are formatted with [[wikilinks]] for key terms to encourage vault interconnection.Use Education when building study notes, course materials, concept explanations, or any content where learning retention and accuracy matter more than brevity.
Vault architecture: structure, indices, and navigation.Agents act as vault architects. Given a project, course, or domain, they immediately produce a full folder tree and create _Index.md Maps of Content (MOCs) that link to all child notes.Naming conventions are applied consistently: PascalCase for folders, kebab-case for notes. The PARA method (Projects / Areas / Resources / Archive) is the default top-level structure unless you specify otherwise. For academic contexts, the default hierarchy is Year > Semester > Course > Unit > Topic. Every generated note includes suggested frontmatter (tags, aliases, date).Use Organization when restructuring your vault, starting a new knowledge base, or any time you need a coherent, navigable folder hierarchy built from scratch.
Senior researcher: maximum academic and technical rigor.Agents apply first-principles reasoning and structure every response as:Context → Analysis → Findings → Limitations → ReferencesResponses target 800+ words with extreme depth and academic precision. This mode is built for tasks where thoroughness and sourcing quality matter — not quick answers.Use Research when investigating a topic in depth, producing literature-style summaries, or any task where you need the rigor of a senior research fellow rather than a general assistant.
Systems architect: planning and blueprints.Agents produce detailed step-by-step execution guides, Gantt-like project breakdowns, and technical specifications including API designs, database schemas, and flowcharts. Responses target 600–800 words and prioritize thoroughness in the planning phase over speed.Use Planning when designing a system, scoping a project, producing a technical spec, or any task where you need a structured blueprint rather than prose.
Extreme mode: maximum synthesis and depth.All other modes applied simultaneously. Agents combine Education’s scaffolding, Organization’s MOC structure, Research’s academic rigor, and Planning’s blueprint methodology in a single response. No response is shorter than 1,000 words. Every angle is exhausted, edge cases are identified, and output targets maximum technical and strategic density.Use Token-Hungry when you need encyclopedic coverage of a topic and are willing to trade response speed for depth. Not recommended for simple tasks.
Start with Standard for most tasks. Switch to a specialized mode when you have a clear need — Education for learning content, Organization for vault restructuring, Research for deep investigations, Planning for system design, and Token-Hungry when you genuinely need exhaustive coverage.