name: raw-document description: "Last-resort skill for working directly with raw OOXML (Office Open XML) or ODF (OpenDocument Format) document structure. Use ONLY when the higher-level skills ($docx-custom, $pptx-custom, $xlsx-custom, $office-custom) have failed, are not applicable, or when the task explicitly requires schema-level, package-level, or cross-format work: validating against XSD/RELAX NG schemas, recovering malformed archives, researching the format specification, mapping elements between OOXML and ODF, or authoring tooling that operates below the library API layer. Do NOT use for ordinary document creation, editing, or extraction — prefer $docx-custom, $pptx-custom, or $xlsx-custom first."

Raw Document Structure

Last resort. Before loading this skill, confirm that none of these cover the task:

• Ordinary .docx work → $docx-custom
• Ordinary .pptx work → $pptx-custom
• Ordinary .xlsx work → $xlsx-custom
• Unpack / repack / validate OOXML → $office-custom

Load this skill when the task requires direct work at the package or XML level that the higher-level skills cannot reach: schema validation, format recovery, specification research, cross-format element mapping, or building tools that parse raw parts.

Intent Router

When to Use This Skill

Appropriate uses

• Schema validation — validating a document part against XSD or RELAX NG schemas in assets/schemas/
• Format recovery — fixing a malformed or truncated archive where a library cannot open the file at all
• Specification research — answering questions about the ECMA-376 / ISO 29500 or OASIS ODF specification at element or attribute level
• Interoperability work — mapping features across OOXML and ODF, diagnosing round-trip conversion losses
• Tooling development — writing parsers, validators, or transformers that operate on raw ZIP parts rather than through a document library
• Edge-case debugging — when a library produces structurally valid but semantically wrong output and the cause must be traced in raw XML

Inappropriate uses (use the other skills instead)

OOXML Quick Reference

Package Structure (.docx / .xlsx / .pptx)

All OOXML files are ZIP archives (OPC packages).

document.docx
├── [Content_Types].xml          ← MIME type for every part
├── _rels/.rels                  ← top-level relationships
├── word/
│   ├── document.xml             ← main body (w: namespace)
│   ├── styles.xml
│   ├── settings.xml
│   ├── numbering.xml
│   ├── theme/theme1.xml
│   ├── media/                   ← embedded images
│   └── _rels/document.xml.rels  ← part relationships
└── docProps/
    ├── app.xml
    └── core.xml

Full layout reference → references/ooxml/03-package-structure.md

Key OOXML Namespaces

Full namespace reference → references/ooxml/04-xml-namespaces.md

Conformance: Strict vs Transitional

ODF Quick Reference

Package Structure (.odt / .ods / .odp)

document.odt
├── mimetype                     ← first file; uncompressed; signals ODF type
├── META-INF/manifest.xml        ← lists every file in package
├── content.xml                  ← document content (text:, table:, draw:)
├── styles.xml                   ← named styles and page layouts
├── meta.xml                     ← document metadata (dc:, meta:)
├── settings.xml                 ← application-specific settings
└── Pictures/                    ← embedded images

Full layout reference → references/odf/03-package-structure.md

Key ODF Namespaces

Full namespace reference → references/odf/04-xml-namespaces.md

Schema Assets

All schemas are in assets/schemas/. Do not fetch them from the network — load from the local copies.

ODF Schemas (RELAX NG)

OOXML XSD Schemas (Transitional, normative per ISO/IEC 29500)

OOXML RELAX NG Schemas (normative per ECMA-376)

Full schema inventory → references/schemas-README.md

Common Raw Operations

Inspect a document's XML parts

# List all parts
unzip -l document.docx

# Inspect a specific part
unzip -p document.docx word/document.xml | xmllint --format -

# Inspect content types
unzip -p document.docx "[Content_Types].xml" | xmllint --format -

Validate against XSD

# Validate WordprocessingML document part
xmllint --schema assets/schemas/ooxml/wml/wml.xsd word/document.xml --noout

# Validate ODF content
xmllint --schema assets/schemas/odf/v1.3-os/OpenDocument-v1.3-schema.rng \
  content.xml --noout --relaxng

Validate with Python (xmlschema)

import xmlschema
import zipfile

schema = xmlschema.XMLSchema("assets/schemas/ooxml/wml/wml.xsd")

with zipfile.ZipFile("document.docx") as z:
    with z.open("word/document.xml") as f:
        errors = list(schema.iter_errors(f))
        for err in errors:
            print(err)

Unpack and repack manually

# Unpack (use office-custom scripts if available)
mkdir unpacked && cd unpacked && unzip ../document.docx

# Repack preserving OOXML requirements (mimetype must be first and uncompressed for ODF)
cd unpacked
zip -r ../output.docx . -x "*.DS_Store"

# For ODF: mimetype must be first and stored (not compressed)
zip -r0 ../output.odt mimetype
zip -r ../output.odt . -x mimetype -x "*.DS_Store"

Relationship to Other Skills

• $office-custom — provides unpack.py, pack.py, validate.py, and soffice.py scripts that handle the raw ZIP round-trip workflow for OOXML. Start there for OOXML unpack/repack tasks.
• $docx-custom — XML editing patterns, tracked changes, comments, and images in Word documents. Use raw-document only when docx-custom's XML Reference section is insufficient.
• $pptx-custom — presentation structure. Refer to references/ooxml/07-presentationml.md here for element-level detail not in the pptx-custom skill.
• $xlsx-custom — workbook structure. Refer to references/ooxml/06-spreadsheetml.md here for element-level detail not in the xlsx-custom skill.