Christian Busse
← All projects

AI & developer tools

Intent & Context Engineering

A practical workflow for turning a vague request into a source-grounded brief an AI system can use.

Toolkit & working paper
Decision-loop diagram connecting observable events, decision state, context policy and bounded actions.
Paper figure. A figure from the project’s working paper, showing how observable events update decision state, shape context and pass through a bounded action gate.Open full-resolution image : Decision-loop diagram connecting observable events, decision state, context policy and bounded actions.

The idea behind the work

Useful AI context needs a clear task, relevant sources, constraints and a way to check the outcome. This project combines a working paper with templates, source-gathering guides and an offline local workflow.

Inside the project

  • Task-specific guides cover project orientation, edits, diagnosis and evidence collection.
  • A Python CLI builds reviewable context packages with source checks and revision history.
  • Browser demonstrations show how a changed scope invalidates old context and requires a refreshed result.

Current scope

A practical toolkit and working paper. Demonstrations use fictional tasks and simulated approval; the software does not require a model provider or API key.