Keep research threads open #

View this page as Markdown

Prompt:

I'm comparing battery suppliers over the next few weeks. Don't lose the open questions between sessions.

Also works: "Keep a running list of what we still need to find out." "What was left open from last week's research?"

What your tool does #

  • Reads the open issues with find at the start of each session, and tells you which threads are open.
  • Raises each open question with raise, kind question, one question per issue, with what is known and where to look next.
  • Raises each piece of work to do as a task, with a due date if you give one.
  • Sets a question that waits on more evidence, such as a supplier's reply, waiting with wait and what would settle it.
  • Comments on an issue with each finding, and closes it with close once the question is answered.

Before your AI starts #

This section is for your AI coding tool: what it checks and gathers before it begins. You don't need to do these steps yourself.

  • The space the project uses: the one .tztrack/config.json gives, or the account's one space.
  • The open questions already raised, read with find, so no thread is raised twice.
  • The area to file the threads under, such as the project's name, so find can list them together.

Steps #

1. Read what is open #

Call find for the open issues in the research's area, and summarize the threads still open for your person.

2. Raise each open question #

Search with find first. If no issue holds the question yet, raise it with the kind question. Give what is known so far and a suggested next step, such as a source to read.

3. Set what waits on evidence #

When a question waits on evidence from outside, such as a supplier's reply, call wait with what would settle it, such as "the supplier's price list". The issue's state becomes waiting, and find with the state waiting lists it. An issue left waiting with no new report for 90 days closes as unexplained, as Issues and kinds says. If the thread should stay open instead, such as one waiting for a report due next quarter, call wait with waiting_for set to outside:<what>. The issue then stays open, and nothing closes it on its own.

4. Add each finding #

Add each finding to its issue with comment, so the next session reads the thread whole with read. When the evidence a waiting question needed arrives, add it, then call reopen with the reason so the issue is open again.

5. Close an answered question #

Close the issue with close, the outcome done, and the answer as its resolution. Then list the issues you raised, by number.

Expected result #

Each open question is its own issue in the research's area, with its findings in its comments. The next session's find answers the threads still open, and those in the state waiting.

Refusals #

This table lists the refusals this task most often meets; Refusals lists every refusal with its cause and remedy.

Refusal Status Cause Remedy
body_too_large 413 The call's text is too long. Shorten the text, or split it into comments.
issue_not_found 404 No issue of the space holds that number. Check the number with find.
space_mismatch 403 The space named is not one your account owns or is a member of. Name a space your account owns or is a member of, or leave space out.