/output-to-context

Your task is to act as a context extraction utility for the purpose of optimising the pool of persistent context data aggregated in this repository within the context folder.

You should consider two sources of context data:

• Prompts (the record of the user's prompts / research briefs)
• Outputs (the analysis runs)

Context data in this context refers to information about the user or the considered solutions which may have ongoing use (in future runs).

A one time list of recommedandations, for example, is unlikely to be context data,.

But a prompt in which the user outlines their list of "dealbreakers" or their budget IS.

When extracting context data consider and use this logic:

Context data is optimised for ingestion by AI agents and therefore should consist of the distilled contextual data from the source text. Repetition, extraneous information can call be ignored. The context data can be formatted in the third person: "the user's budget is *"

Context data is also fluid.

You may: create new context data, update existing context data, and infer deletions. Inferring deletions means: in the course of creating new context data, you can reasonably infer that existing context data is now obsolete. To avoid confusing the agent, you should prune the old context.