SuperAPI AI Directory
Partner with us中文AI newsFavorites
Skillslaw-to-markdown

law-to-markdown

Writing & research

View repository

Updated 10-16 07:47

About this skill

Legal Work Master Control Rules (Mandatory)

Before executing this Skill, you must first follow:

  • skills/legal/法律工作总控/references/practice-profile.md
  • skills/legal/法律工作总控/references/matter-workspace-protocol.md
  • skills/legal/法律工作总控/references/document-reading-protocol.md
  • skills/legal/法律工作总控/references/source-boundary-protocol.md
  • skills/legal/法律工作总控/references/ocr-correction-protocol.md
  • skills/legal/法律工作总控/references/pkulaw-mcp-legal-verification-protocol.md

This Skill only handles tasks within this specialty; case isolation, source disclosure, material gaps, statute/Wiki verification, OCR correction, and reading review are executed according to the master control rules.

Repeal of Old Rules (Mandatory)

  • Client directories, stage directories, old-style ledger writes, and old local reading protocols hard-coded directly in old texts are not to be used as execution rules.
  • Matter paths, current matters, system records, business file areas, and review ledgers are uniformly based on the Legal Work Master Control matter-workspace-protocol.md .
  • Do not silently write to the review ledger; when an update is truly needed, first confirm that it belongs to a review ledger update and explain it to the user.
Law To Markdown

Processing Rules

  1. Input is .txt: directly save as .md。
  2. Input is .pdf / .docx：
    • First check whether mineru-ocr skill。
    • is installed. Not installed: prompt to install mineru-ocrfirst, installation address: https://github.com/cat-xierluo/legal-skills/tree/main/skills/mineru-ocr
    • Installed: prioritize calling mineru-ocr for processing.
    • Call failed: first prompt to check mineru-ocr configuration/Token.
    • Only when the user explicitly agrees, use the local fallback (python-docx / pdfplumber）。
  3. After the first stage is completed, by default execute the second stage format adjustment (format only, do not change the original text characters):
    • The large model calling this skill first determines "legal/non-legal" and passes the result to the script.
    • If the caller does not pass a result, the script uses hard rules for automatic recognition (--law-decision auto）。
    • Law name #
    • Part/Sub-part ##
    • Chapter ###
    • Section ####
    • Article #####(only "Article X" is a heading;第X条【条标】a whole line as a heading is not split)
    • Paragraph/Item/Sub-item have no headings
    • Items and sub-items are line-broken by marker
    • Clean up excess spaces: remove trailing spaces, remove leading ASCII spaces, remove full-width indentation at the start of body lines, normalize spaces after headings
  4. If recognized as clearly non-legal text (such as GB/standard-type documents), the second stage explicitly refuses:Stage2: rejected (non-law-document)。
  5. If the second stage does not recognize legal structure or the fidelity check fails, automatically no-op, without changing the text.
  6. By default execute the third stage check (two sub-stages, hard threshold):
    • Stage3-A: verify stage2 relative to stage1 textual content accuracy (the character stream must be identical after removing heading symbols and whitespace)
    • Stage3-B: verify structural effect (structural hierarchy, article heading rules, non-legal strategy, space conventions, item/sub-item line breaks)
    • Any failure triggers an automatic retry, up to 2 times; if it still fails, it exits with an error by default

Stage 2 recognition parameters

  • --law-decision law: The caller has determined it is legal text, so optimize directly according to the legal structure.
  • --law-decision non-law: The caller has determined it is non-legal text, so Stage 2 rejects it directly.
  • --law-decision auto: The default mode when no determination is passed, using hardcoded rules in the script.

Stage 3 parameters

  • --skip-stage3-check: Skip the Stage 3 check (not skipped by default).
  • --stage3-max-retries: Number of automatic retries after failure, default 2。
  • --stage3-strict / --no-stage3-strict：
    • Default strict mode (exit with an error on failure)
    • Non-strict mode only outputs a report and does not block the process
  • --artifact-level minimal|standard|debug：
    • minimal(default): Delivery-oriented, minimal output
    • standard: Keep process files (stage1/stage2/stage3-check) for troubleshooting
    • debug: Keep all process artifacts (including debug information)

Output rules

  • By default, output to the markdown/ subdirectory in the same directory as the input file, and create a separate directory based on the input file name:
    • markdown/<文件名>/
  • Default (--artifact-level minimal) output:
    • <原文件名>+审核报告.md(detailed process and conclusion)
    • <原文件名>+最终成果.md(generated only for legal text that passes review)
  • Non-legal documents (such as GB/standards):
    • The conclusion is "reject processing"
    • Only output <原文件名>+审核报告.md
    • Do not output <原文件名>+最终成果.md
  • When review fails, process files are automatically retained for troubleshooting:
    • <文件名>.stage1.md
    • <文件名>.stage2.md
    • <文件名>.stage3-check.md
  • If more artifacts are needed, you can switch:
    • --artifact-level standard: Keep stage1/stage2/stage3-check
    • --artifact-level debug: Keep all debug artifacts
  • You can specify via command parameters --out or --out-dir。

Common commands

python3 law-to-markdown/scripts/law_to_markdown.py "input.txt"
python3 law-to-markdown/scripts/law_to_markdown.py "input.docx"
python3 law-to-markdown/scripts/law_to_markdown.py "input.pdf"

When the user explicitly agrees to fallback:

python3 law-to-markdown/scripts/law_to_markdown.py "input.docx" --allow-fallback
python3 law-to-markdown/scripts/law_to_markdown.py "input.pdf" --allow-fallback

License

SPDX-License-Identifier: Apache-2.0

Copyright (c) 2026 JiCheng

Licensed under the Apache License 2.0; the original license text is in the repository root directory LICENSE。