PaperBD Study Paper

Guide the user from paper identification to answering questions in Paper Breakdown.

1. Identify the paper

First, find the paper's arXiv ID.

If the user already gave a valid arXiv ID, use it.
If the user gave only a title, author, or vague description, first do a web search to find the arXiv ID.
If the web search does not produce a clear arXiv ID, search export.arxiv.org second.
Never guess or infer the arXiv ID from partial information.
Never spend excessive effort trying to recover the arXiv ID. If the web search and export.arxiv.org search both fail, ask the user to provide the arXiv ID directly.
If multiple papers could match, ask one targeted follow-up question or ask the user to provide the arXiv ID directly.
Do not proceed to PaperBD commands until you have a specific arXiv ID.

Prefer paperbd status.
If the CLI indicates the user is not logged in, tell them to run paperbd login.
Explain that login completes in the browser and then returns to the terminal.
Stop and wait for confirmation before continuing.

paperbd usage returns:
- remaining paper analyses
  - Paper analysis means accessing new papers the user has never read with Paper Breakdown.
- remaining CLI requests
  - CLI requests means every successful CLI query reduces this count.
- when the current quota resets
paperbd papers returns the list of papers the user currently has access to.
paperbd ask --arxiv <arxiv_id> --query "<question>" automatically tries to add the paper to that list.
If you have access to the terminal, run these commands yourself. If not, ask the user to run them.
Even if remaining paper analysis is 0, still try paperbd ask.
If you want to verify whether the user already has access to a paper, check paperbd papers | grep <arxiv_id>.
If paperbd ask succeeds, continue with the answer.
If paperbd ask fails because the paper is not accessible, explain the error and next action.

Run the following in the terminal:

paperbd ask --arxiv <arxiv_id> --query "<question>"

paperbd ask returns actual PDF passages relevant to the question.
paperbd ask also automatically tries to add the paper to the user's accessible paper list.
If you have access to the terminal, run this command yourself. If not, ask the user to run it.
This will print passages directly into the terminal
Ask simple questions

After the paper is identified and access is confirmed, answer the user's actual question about the paper.

If the user has not asked a concrete question yet, ask one targeted next question.
Ask targeted questions one at a time.
Keep follow-ups specific.

401 Unauthorized
- Auth issue. Ask user to paperbd login first
400 Bad Request
- Missing one query, or arxiv_id
429 Too Many Requests
- The user exceeded the CLI rate limit: 5 requests per 60 seconds
- Tell them to wait and follow Retry-After if present
Fetch failed
- This is likely a sandbox network permission error.
- Ask the user for permission to make network requests with paperbd, then try again.
403 Forbidden
- FREE tier cannot use CLI retrieval
- Or monthly CLI quota was exceeded: STUDENT = 100, RESEARCHER = 200
404 Not Found
- The paper is not accessible to that user
- Tell the user to start the paper analysis from https://paperbreakdown.com/search?query=<arxiv_id>
500 Internal Server Error
- Unexpected server-side failure, such as retrieval pipeline or backend issues

When explaining an error, state the likely cause and the next action. Do not invent CLI capabilities.

Be operational and direct.
Move the user to the next blocking step quickly.
When blocked on installation, login, or access, do not continue as if the paper is available.
Prefer one focused follow-up question over a large questionnaire.