v0.3.0

Added

• New get_judgment_pdf tool — returns or downloads a judgment's PDF. Behavior controlled by the save_to arg or MCP_TW_JUDGMENT_DOWNLOAD_DIR env var; when neither is set, only the URL is returned. Re-calls with the same path return cached: true without re-downloading.
• get_judgment now returns a paragraphs field alongside content — a flat list of {id, section, level, heading, text} entries with stable hierarchical IDs (e.g. 理由.一.(三).2) for precise citation and downstream highlighting.

Changed

• parser/judgment_parser.py gains extract_pdf_url and parse_paragraphs; parse_detail now emits paragraphs (additive; content preserved for backward compatibility).
• connectors/rest_client.py gains api_get_bytes for binary downloads.

Install

uvx mcp-tw-judgment
# or
pip install mcp-tw-judgment==0.3.0

Full Changelog: v0.2.0...v0.3.0