Excellent! Ok now, I'd like to develop another test case that is a little different and also very real. I need to get my resume in order for a job listing at the company called Everlaw. My friend Leo told me about it and as an internal reference I just need to get Leo my resume to pass along. So I need help updating (and totally changing/recreating) my resume accordingly. To that end, I have added the project folder for this in projects/resume_for_everlaw off root. In that directory you will find sevel (7) files. Please do the following:

1. read each file and give me a blurb of the FULL CONTENT of each file, organize by filename, so I know you've got all the relevant info in active attention. Note that a lot of the deep context is found in the transcript of the call with me and Leo and with the actual job descriotion and my current draft resume but the other files also all have relevant info.

2. come up with a plan to create a new workflow for this, which must conclude with at least one finished and fit-for-purpose resume and a change log capturing all the major decisions that took us from the curent resume version to the final version(s), and may possibly include a new design pattern, new schemas, and new promots, as well as some new ways to ensure content from multiple files is ingested and passed forward at least for the initial analysis during the first round.

3. Note that some of the files are in PDF form and this may require sending PDFs as kind of attachment to Claude, GPT, and Gemini models. Therefore you may use the documentation for that, including the relevant specifications for each API, that I put together for you in archive/attach-files-to-api-calls.md off the root. Include your plan for sending the PDF info to the models for at least the first round, whether that be to attach it or to transform it first on our end or otherwise. Whateve you think is best. But keep in mind this process of having the user include varied filetypes of attachments wil become routine and the typical way new Triple Helix project are started, in order to avail the syste of all the relevant context. So we will want a solid, simple, easy to use and reference and rely on method for this. Sending along with the API calls seems basic to me but then again, I am not that technical so, again, you figure this part out please and do a good job for this workflow and also for all future instances of yoursef and future workflows we aim to support.

4. Note that I think the right way to do this type of project is first to identify and extract all the key info from the initial context according to what is relevant based on the user's goal and any instructions the user provides and THEN to start pushing the info through a few rounds to generate, select, revise, and finalize just the right deliverable but I am not sure, Please think hard about the work flow, perhaps suggest options, and you can consider my own initial approach (though I don't necessarily recommend it) found in projects/WorkingSessions/2025-11-13-WorkingSession.md with Request 1 (on line 1) and Request 2 (on line 439) and the respective corresponding responses from each model as one way to approach eliciting the relevant information and some good starting candidates for the final deliverable (namely one or more well crafted resume version fit for purpose to send to Leo and be successful to garner a job offer from Everlaw).