Sloth-ninja gives JessicaOSS an English-law working wardrobe

Applies the 16 MECHANICAL rows from the terminology audit report: UK spellings
(analyse/analysing, summarise) in chat placeholders, tabular column-preset
prompts, and workflow templates; en-GB date locale in the credits-exhausted
modal so reset dates render as "8 July 2026" rather than US month-day-year.

No legal-terminology judgment calls in this commit - those are handled
separately (Disclosure Review rename) or left for owner sign-off per the audit
report's LEGAL TERM OF ART / UNCERTAIN classifications.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

Part A (remaining 2 rows for this file): UK spelling fixes in the Change of
Control Review column prompts (summarise).

Part B (DECIDED, MIGRATION_SPEC §6.1): renames "E-Discovery Review" to
"Disclosure Review", reframes it around CPR Part 31 / PD 57AD, and splits the
single "Privileged?" column into three: Legal Advice Privilege, Litigation
Privilege, and Without Prejudice, each Yes/No + basis with uncertainty noted
explicitly. The id is deliberately kept as "builtin-ediscovery" rather than
renamed: backend/schema.sql's hidden_workflows.workflow_id is a free-text
column keyed on this exact string, so renaming would silently un-hide the
workflow for any user who had previously hidden it (see code comment at the
template definition, and docs/BUILD_LOG.md for the full rationale).

Part C adds five new UK workflow templates: English-law SPA Review (tabular,
Corporate), Commercial Lease Review - LTA 1954 (tabular, Real Estate, with a
dedicated security-of-tenure / s.38A contracting-out column and a 1954 Act
red-flags column), TUPE Analysis (assistant, Employment, structured reg
3/4/7/13-15 analysis), Employment Contract vs Statutory Minima (tabular,
Employment, contractual term vs statutory floor per column), and Companies
House Due-Diligence Snapshot (assistant, Corporate, names the CH tools and
instructs the model not to invent register data when they're unavailable).

All statutory references used in these templates are real, resolvable
citations (verified live against legislation.gov.uk - see evals/cases/cit-*
added in a following commit). No case-law citations anywhere in the new
content, per the Find Case Law deferral.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

Adds 22 new eval case files (17 citation, 5 judged) and 5 new rubric files,
covering all five new WS4 templates plus the renamed Disclosure Review (which
gets its own judged rubric per the three-way privilege split); the Companies
House Due-Diligence Snapshot template reuses the existing
jud-due-diligence-snapshot.yaml + rubric per the brief and is extended with 2
new citation cases rather than duplicated.

Every citation case's input.text is a short realistic excerpt of what the
matching template instructs the model to produce, using the exact statutory
forms the templates mandate (s.N Act Year / SI year/number). None are marked
smoke (the smoke subset is already at its 5-case cap). Judged cases are marked
pending (no workflow-fixture output yet) with real rubrics ready for when
fixtures land.

Verified: `npm run evals` from the worktree root - 20 passed, 0 failed, 1
skipped (Companies House deterministic case, no API key set - expected), 6
pending (all judged cases). Every new citation case resolved live against
legislation.gov.uk with zero failures. `npm run evals:smoke` unchanged (3
passed, 1 skipped).

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

Records scope, key changes, verification evidence, and the for-owner-review
list of legal-terminology judgment calls made in new template content. Notes
the expected trivial merge conflict with parallel WS1-WS3 branches also
inserting at the top of the file.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

Signed off 8 July 2026: keep 'lawyers' (persona); opinion→judgment in the
core prompt; add SONIA alongside SOFR/EURIBOR/base rate; E&W company
example; UK-first examples everywhere (English Law first, GBP, en-GB
dates pinned in five browser-locale spots, four-weekly pay frequency);
waterfall wording leads with the descriptive term, both market labels
kept. docs/LEGAL_LANGUAGE_REVIEW.md logs all agent legal-language
judgments for solicitor review.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

Due-diligence snapshot (live register data, 13927967) and TUPE analysis
(synthetic scenario) committed as provisional input.text - interim Opus
judge pass scored both 5/5; cases skip until ANTHROPIC_API_KEY exists,
then run automatically. Inputs to be reviewed with an AGL associate.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>