The Token-Accounting Bug Report is out
We audited ~110 AI usage/cost tools with synthetic fixtures and filed 65+ evidence-backed findings — 10 already merged/accepted by maintainainers. The five classes that keep appearing (per-block re-summation, re-emitted events, cache-pricing semantics, price-table drift, resume/fork loss) are documented with links to every public filing:
📄 Read the report
Maintain a usage tool? Audit it in 10 minutes with the open fixture corpus (plain JSONL, no runtime install).
If the report or the fixtures saved you a bug, a ⭐ helps others find them — and a link to your fixed issue grows the shared casebook.
The Token-Accounting Bug Report is out
We audited ~110 AI usage/cost tools with synthetic fixtures and filed 65+ evidence-backed findings — 10 already merged/accepted by maintainainers. The five classes that keep appearing (per-block re-summation, re-emitted events, cache-pricing semantics, price-table drift, resume/fork loss) are documented with links to every public filing:
📄 Read the report
Maintain a usage tool? Audit it in 10 minutes with the open fixture corpus (plain JSONL, no runtime install).
If the report or the fixtures saved you a bug, a ⭐ helps others find them — and a link to your fixed issue grows the shared casebook.