The learning evidence loop
How a Project 42 lesson becomes practice, feedback, a transcript entry, and eventually a badge.
Explore this visual →Visual guides
See the system, not just the steps.
Explore provider-neutral diagrams for learning, research, prompting, tool safety, agent workflows, and trustworthy content updates. Every visual includes a plain-language explanation and editable Mermaid source.
Source first · Accessible by design
8 visual guides
Static SVG in the browser. Mermaid source in the repository.
A grounded answer workflow
A repeatable path from a real question to an evidence-linked answer with uncertainty made visible.
Explore this visual →
The bounded agent loop
A safe agent cycle that separates planning, authorization, action, verification, and terminal states.
Explore this visual →
An effective prompt contract
The layers that turn a vague request into an auditable task contract across providers.
Explore this visual →
Choose an AI provider by evidence
A neutral decision process that tests real workloads instead of choosing from marketing claims.
Explore this visual →
Tool and data trust boundaries
Where identity, policy, approval, validation, and audit controls belong around an AI tool call.
Explore this visual →
A verifiable multi-agent handoff
How scoped context and evidence move between an orchestrator, worker, independent reviewer, and human.
Explore this visual →
From source drift to a trusted release
The human-gated workflow for detecting changed AI guidance and releasing evidence-linked updates safely.
Explore this visual →