same-page

Align goals and understandings with the user; get on the same page about something

SKILL.md

GitHub

Create baseline understanding.

Use primary sources to compile your current understanding of the query.

Produce a scratch file containing:

  1. A baseline summary of your understanding, in narrative form.

Use the southpark rule between, relationships between narrative details should be "x therfore y or a however b". Rarely will this be a story, so it's fine to organize sub-narratives which are less connected.

This should naturally lead to 2. a heirarchy of goals and desired outcomes

Which should also have their own related.

  1. challenges.

Align on a core principle/desire

Identy the core desire/goal and present it to the user, along with brief justification, and an abridged version of your baseline narrative to provide enough context to be on the same page.

Make sure your presentation hits obvious edges where your understanding and the users understanding can be misaligned:

  1. Direction - where are we before achieving/addressing, where are we after?
  2. Reason - Why do we care to move in that direction?
  3. Scope - How far do we want to push with this goal?

If applicable further clarify scope by highlighting 1-2 challenges subproblems, Nominating one of the simplest to solve, and one of the most complex to solve.

Iterate with the user untill you agree on a top level goal, proposing refinements, drawing from other high level goals you identified, or generating goals de novo as needed.

Beyond iterating proposals you should solicit feedback, and visit primary sources as necessary.

Goals outside the purview of the core goal.

For remaining goals that don't fall under the core goal/desire repeat the goal alignment process.

using the scratch file

The scratch file is to make sure you/the user don't get lost in the weeds.

Such discussions are want to twist and fork, and it's your job to keep log of narrative and goals to make sure things aren't lost in the sauce.

user shorthands

y - yes/agree n - no/disagree dc - don't care, I think this decision isn't super important