Run2 pdf form filling

Skill cxcscmu/SkillLearnBench/skills/b2-self-feedback-claude-opus-4-6/court-form-filling/run2_pdf-form-filling

[COLM'26] SkillLearnBench is the first benchmark for evaluating continual learning methods that automatically generate agent skills.

Install

npx -y skills add cxcscmu/SkillLearnBench --skill run2_pdf-form-filling

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

What its author says it does

Copied from the file, not written here

Complete workflow for filling fillable PDF forms using Claude Code PDF skill scripts with pypdf.

SKILL.md

2.7 KB, as published. Nobody here has run it

Filling Fillable PDF Forms

Prerequisites

Python 3 with pypdf installed
Scripts located at /root/.claude/skills/pdf/scripts/
Always run scripts from: cd /root/.claude/skills/pdf

Step-by-Step Workflow

1. Check if PDF has fillable fields

python3 scripts/check_fillable_fields.py <input.pdf>

If fillable, continue below. If not, use the non-fillable annotation workflow in forms.md.

2. Extract field metadata

python3 scripts/extract_form_field_info.py <input.pdf> <field_info.json>

Output JSON contains field_id, type, page, rect, and for checkboxes: checked_value/unchecked_value.

3. Convert PDF to images for visual mapping

mkdir -p <output_dir>
python3 scripts/convert_pdf_to_images.py <input.pdf> <output_dir>

Important: Create the output directory first or the script fails.

4. Map fields to visual form elements

Read the generated images to understand what each field_id corresponds to
Use field names as hints (e.g., PlaintiffName1, DefendantAddress1)
Use bounding box rect coordinates to locate fields on the page

5. Create field_values.json

[
  {
    "field_id": "exact.field.id.from.extract",
    "description": "Human-readable description",
    "page": 1,
    "value": "text value"
  },
  {
    "field_id": "some.checkbox.field",
    "description": "Check this box",
    "page": 2,
    "value": "/1"
  }
]

Value rules:

Text fields: Plain string value
Checkboxes: Use checked_value from field_info.json to check (e.g., "/1", "/On"), unchecked_value to uncheck (e.g., "/Off")
Radio groups: Use one of the value entries from radio_options
Omitted fields: Left blank/unchanged — only include fields you want to fill

6. Fill the form

python3 scripts/fill_fillable_fields.py <input.pdf> <field_values.json> <output.pdf>

Script validates field IDs and values. Fix any error messages and retry.

7. Verify output

Convert filled PDF to images and visually inspect each page:

mkdir -p <verify_dir>
python3 scripts/convert_pdf_to_images.py <output.pdf> <verify_dir>

Common Pitfalls

Missing output directory: convert_pdf_to_images.py doesn't create directories automatically
Wrong field_id: Must be exact match from extract_form_field_info.py output
Wrong page number: Must match the page value in field_info.json
Checkbox values: Don't use "true"/"false" — use the exact checked_value/unchecked_value strings
Script directory: Must cd to /root/.claude/skills/pdf before running scripts