The collection / Working preview

Context for
a specific question.

Choose a task. Inspect the evidence.
Prepare context for your AI workflow.

Explore the collection ↓
01 / Locate the work
Choose where the work happens.Drag to turn · Select a place to explore

Open a tool to inspect its knowledge, sources, method and graph.

Opening the collection…

63 context frameworks · 42 artifact examples across 21 workflows. Readiness and missing case inputs are marked on each object.

02 / Put knowledge to work

Prepare the AI input.

Choose tools, describe your task and inspect the evidence that enters the AI input.

Sources→Expert framework→Your task→Artifact

Choose up to six tools. Sources and review limits stay attached.

What the AI receives

Your task, its evidence,
and the checks that matter.

How does this connect to an agent?

The desk reads the selected objects from the database, selects complete knowledge sections within a byte budget and returns cited context. The same preparation function is exposed through a read-only MCP adapter for compatible clients. Model execution and Praxis integration are separate work in progress.

Technology and research notes ↗

Saliency means keeping the distinctions that matter for this task. Efficiency means spending less context on irrelevant material. Model-quality gains require task-level evaluation; the desk does not claim them.

A connected knowledge base

Inspect the knowledge
behind the tools.

The collection connects to 20 semantic layers, 13 ontology objects and a dated global source directory.

Explore the knowledge structures ↗
Object / Inspect before use

Source inspection