v0.1.1

• The repository is English. Shop tickets, templates, the FAQ reply, sanitizer markers, the keyword lexicon, fixtures, and docs are English. The small-sample rate label is not conclusive. The fictional shop name is Brujula Store.
• Live accuracy and cost from 2026-10-17 stay as measured on the previous bilingual corpus. They were not re-run. Simulated profile figures were remeasured on the English corpus on 2026-10-23 (python -m triage_lab eval, commit 4e32b1c1bfade248a9b747f0feb3fa0dbe5ee23c). large_sim test accuracy stays 0.9655. Test human share is 0.1724 at tau 0.80. Blended cost is 0.174 USD per ticket [EST]. Saving versus all-human is 0.826 USD per ticket [EST].
• Security reports go through GitHub private vulnerability reporting: open the Security tab and choose "Report a vulnerability".
• docs/HOW_IT_WORKS.md is a walkthrough for engineers and buyers. Interview-script framing is removed from that page and from the README docs index.
• .gitleaksignore lists the gitleaks fingerprints for the secretscan.py sha256 line in docs/PROVENANCE.md. That line is a digest. Gitleaks labels it generic-api-key.
• Translated the remaining Spanish documentation into English: docs/HOW_IT_WORKS.md, and leftover phrases in docs/BUSINESS_CASE.md, docs/COST.md, and docs/LIVE_RUN_TEMPLATE.md. Numbers, commands, links, and the labels [EST], [SYN], MEASURED, SIMULATED, and UNVERIFIED stay. The small-sample rate label is not conclusive.
• Expanded the English README to the same shape as the other portfolio labs: a three-line start, quickstart, use cases and governance, a short business-value section, a proof line, and what the lab does not prove.
• Added docs/BUSINESS_VALUE.md. Money figures are the ones already in docs/BUSINESS_CASE.md and docs/COST.md. No new rates.
• Refreshed docs/HOW_IT_WORKS.md so the walkthrough matches the gate, the demo, and the 2026-10-17 live record.
• Noted in SECURITY.md that CI does not enable live mode and does not build the image.