Home

5 February Tasks

1. Add a "working" indicator when waiting for a response from ChatGPT.
2. If code is highlighted, send highlighted code to the LLM.
3. Add option to mark LLM responses as "good" (something to be kept for use in quizzes)
4. Add option to mark LLM response as "training". Training responses should be sent with subsequent LLM queries to help show the LLM what types of questions to generate.

---

22 January Tasks

1. Adjust handlebars template so there are newlines and blank lines where necessary. (By Friday if possible).
   • Render a few questions and make sure you don't see anything obviously wrong.
2. Make caching local instead of global. (Local only is fine for now.)
3. Explore AI (as time allows)

Merge to master when done with 1 and 2.

---

15 January Priorities for the Week

1. Turn off handlebars formatting in Husky commit hook
   • Are there any other commit hooks you suggest turning off?
2. Adjust handlebars template so there are newlines and blank lines where necessary. (approx 3 hours for 1 and 2)
3. Continue to push on allowing lines to be inserted (5 hours max)
4. Explore AI ideas (approx. 2 hours)

---

7 December

• Something (a commit hook, perhaps) is automatically formatting the handlebars template. We need to turn that off.

---

• How difficult would it be to load the categories from a project-specific file?

  • It could be a regular stand-alone file (e.g., qlc-categories.json)
  • Or does it make more sense to get the categories from a vscode project file?

• When leaving comments in an .ipynb file, I think the line numbers are relative to the block, not the entire file. Is there a way to include the block number in addition to the line number?

• Go to View -> Open View -> Comment Explorer. When you do this and select a comment, the "raw" view of the .ipynb file opens. Is there a way to change this behavior?

• How easy would it be to add additional fields? (I'm not sure whether that is necessary, I'm just curious what your first impression is.)