v2.1.1

Full Changelog: v2.1.0...v2.1.1

v2.1.0 — Petition for Rehearing Support

What's New

• Rule 40 support: Bundled N.D.R.App.P. Rule 40 (Petition for Rehearing)
• New brief type: petition_rehearing — auto-detected from cover page text
• Mechanical checks: 10-page limit (PG-005, Rule 40(b)), advisory cover color note, oral argument marked N/A
• Semantic checks: RHR-001 (points overlooked/misapprehended), RHR-002 (applicable Rule 28(b) items), RHR-003 (supporting argument)
• MIT license added

Rule 40 Check Summary

When a petition for rehearing is detected:

• All Rule 32 formatting checks apply (margins, fonts, spacing, paper size, page numbering)
• 10-page limit per Rule 40(b)
• Three new content checks per Rule 40(a)(2) and 40(b)
• Universal semantic checks apply (TOC, TOA, party naming, privacy, citations)
• Appellant/appellee-specific checks are excluded

v2.0.0

What's New

• Rule freshness checking against ndcourts.gov — The skill now automatically verifies that bundled rules are current by comparing effective dates against the live versions on ndcourts.gov. Results are cached for 90 days so most runs make no network calls. Fails open if the site is unreachable.
• Standalone freshness script — scripts/check_rule_freshness.py forces a live check bypassing the cache, useful for manual verification.
• Update checker — check_update.py provides a cached weekly check for new releases.

All 9 bundled rules verified current as of 2026-03-07.

v1.8.0 — Rules 14, 21, and 11.6

What's New

Adds three new rules and 10 new compliance checks:

N.D.R.App.P. 14 — Identity Protection

• PRV-002 through PRV-006: Semantic checks for mental health respondents, guardianship/conservatorship, juvenile respondents, TPR proceedings, and sexual offense victims — all must use initials only.

N.D.R.App.P. 21 — Writs

• WRT-001: Writ petition must state relief sought, issues, facts, and reasons.
• WRT-002: Writ petition must include supporting documents.
• WRT-003: Exhibit citations should use (E{page}:{line/para}) format.

N.D.R.Ct. 11.6 — Medium-Neutral Citations

• CIT-001 (mechanical): Flags N.W.2d citations lacking a nearby YYYY ND ## medium-neutral cite.
• CIT-002 (semantic): Evaluates pre-1997 vs post-1997 citation distinction.

Other

• Fixed _rule_link() to generate correct ndcourts.gov URLs for N.D.R.Ct. rules (3.4, 11.6).
• PRV-001 now cross-references Rule 14(a)(5) overlap.

v1.7.0

What's New

• Cross-platform skill packaging: Added install.sh (macOS/Linux) and install.ps1 (Windows) installers that create a fresh venv at deploy time
• Platform detection in SKILL.md: Commands now use $VENV_PYTHON instead of hardcoded python3, with automatic detection of Windows vs macOS/Linux venv paths
• Portable temp directory: config.py uses tempfile.gettempdir() instead of hardcoded /tmp
• deploy_skill.py: No longer copies .venv (venv is created by install scripts instead)

v1.6.1: Rename to JetBriefCheck

Changes

• Rename project to JetBriefCheck throughout (code, config, CI, docs, skill metadata)
• Update report footer: Generated by JetBriefCheck v1.6.1 (2026-02-25) · Rules current as of 2026-02-17
• Move ZIP to release assets instead of tracking in repo (eliminates push races from automated ZIP rebuilds)
• Fix pre-existing test failures (version 1.5.0 → 1.6.0 mismatch)

Migration

• Re-deploy skill: python deploy_skill.py (new skill path: ~/.claude/skills/jetbriefcheck/)

v1.6.0

Changes

• Fix FMT-009 false positive: Spacing detector now measures inter-block baseline gaps, correctly detecting double spacing in PDFs that encode each visual line as a separate text block. Previously only measured intra-block gaps, causing nearly all briefs to be flagged as single-spaced.
• Page references in report: Failed check cards now show page number badges identifying which pages have issues. Passed checks table includes a Pages column.
• Engine footer: Report footer now indicates whether PyMuPDF was used or fallback mode was taken.
• Data model: Added pages field to CheckResult and pymupdf_used field to ComplianceReport.
• CLI: Added --pymupdf/--no-pymupdf flags to build_report.py.

v1.5.0

Self-contained SKILL.md with fallback mode

SKILL.md is now a single self-contained file (~47KB) that works in both Claude Code and claude.ai projects without needing external rule files.

Changes

• Bundled rules and check definitions — All six ND appellate rule files and full check definitions are appended as inline appendices. No external file reads required.
• PyMuPDF fallback mode — When PyMuPDF is not available (e.g., in claude.ai projects), the skill gracefully falls back to semantic-only checks. Claude reads the PDF directly and produces a text report noting that mechanical checks were skipped.
• Updated workflow — Phase 1 now branches: success → full mode (mechanical + semantic + HTML report); failure → fallback mode (semantic-only, text report).