claude-code - 💡(How to fix) Fix Ordered-list markdown in submitted messages gets auto-renumbered (e.g. '2.'/'5.' → '1.'/'2.'), clobbering intended numbering

Official PRs (…)
ON THIS PAGE

Recommended Tools

×6

Utilities matched from this issue’s tags and category — try them while you read without losing context.

GitHub issue graph ai analysis

Paste a GitHub issue URL. We fetch that issue, discover linked issues from bodies/comments/timeline, collect linked pull requests, and produce a structured English report.

The report is written in English Markdown for sharing and archival.

Helpful · Quick feedback

Loading…

When a user message contains a markdown ordered list with non-sequential or intentionally-specific numbering (e.g. items labelled 2. and 5. to refer back to a prior enumerated list), the rendered/submitted message re-numbers them sequentially (1., 2., …). The original numbering is preserved in the editor but rewritten after send. The user cannot assert "I really do mean 2. and 5." through to the rendered transcript.

Root Cause

When a user message contains a markdown ordered list with non-sequential or intentionally-specific numbering (e.g. items labelled 2. and 5. to refer back to a prior enumerated list), the rendered/submitted message re-numbers them sequentially (1., 2., …). The original numbering is preserved in the editor but rewritten after send. The user cannot assert "I really do mean 2. and 5." through to the rendered transcript.

Fix Action

Fix / Workaround

Workaround the user found

Code Example

2. (responding to your point 2)
   5. (responding to your point 5)
RAW_BUFFERClick to expand / collapse

Summary

When a user message contains a markdown ordered list with non-sequential or intentionally-specific numbering (e.g. items labelled 2. and 5. to refer back to a prior enumerated list), the rendered/submitted message re-numbers them sequentially (1., 2., …). The original numbering is preserved in the editor but rewritten after send. The user cannot assert "I really do mean 2. and 5." through to the rendered transcript.

Repro

  1. In the input editor, type a list using deliberate numbers, e.g.:
    2. (responding to your point 2)
    5. (responding to your point 5)
  2. Confirm it looks right in the editor.
  3. Send.
  4. Expected: the message renders with 2. and 5. as typed.
  5. Actual: the renderer normalizes to 1., 2. — the intended cross-reference numbering is lost.

Why it bites (not purely cosmetic)

In long technical back-and-forths, both sides use numbered lists to track threads ("addressing your 2 and 5"). Auto-renumbering silently breaks that correspondence — the recipient (human or model) can no longer tell which prior point each item answers. It got noticeably worse in one long session after an auto-archive event, where the numbering correspondence between turns desynced entirely.

Workaround the user found

Lettered lists (a., b., c.) appear not to be auto-renumbered the same way — so switching ordered→lettered preserves intent. Mentioned in case it narrows the cause (ordered-list normalization specifically).

Severity

Niggle / quality-of-life. Not blocking, but a recurring papercut in numbered-point discussions. Filed at a user's request ("for giggles") after it tripped a long design conversation.

Vote matrix · Quick signals

Works
Did the solution work? Tap to confirm.
Easy Fix
Was it a quick fix?
Time Saver
Did it save you time?
Blocking
Was it severely blocking?
Common Issue
Are others likely hitting this too?
Flaky / Intermittent
Is it intermittent?
Verified / Reproducible
Can you reproduce it reliably?
Loading…

Still need to ship something?

×6

Another batch ranked right after the header list — different links, same matching logic.

Back to top recommendations

TRENDING

claude-code - 💡(How to fix) Fix Ordered-list markdown in submitted messages gets auto-renumbered (e.g. '2.'/'5.' → '1.'/'2.'), clobbering intended numbering