name: docx-automation description: | Generate DOCX documents from pipeline results or structured data. Uses python-docx library for Word document generation. Independent utility skill (can be called standalone or post-pipeline). user-invocable: true context: fork model: opus version: "3.1.0" argument-hint: "generate <source> | from-json <data.json> | report <image_id> | --workload <slug>" allowed-tools:

• Read
• Bash
• Write
• Task
• mcp__sequential-thinking__sequentialthinking hooks: Setup:
  • type: command command: "source /home/palantir/.claude/skills/shared/workload-files.sh" timeout: 5000

EFL Pattern Configuration (Utility Skill - Minimal)

agent_delegation: enabled: false reason: "Utility skill - direct execution for document generation"

parallel_agent_config: enabled: false reason: "Sequential document generation preferred for consistency"

internal_validation: enabled: true checks: - "Source data is readable and valid" - "python-docx library is installed" - "Output directory is writable" max_retries: 2

output_paths: l1: ".agent/prompts/{slug}/docx-automation/l1_summary.yaml" l2: ".agent/prompts/{slug}/docx-automation/l2_index.md"

/docx-automation - DOCX Document Generation

Version: 3.1.0 | Type: Independent Utility Skill Role: Generate Word documents from pipeline results or data Pipeline: Standalone or post-pipeline utility

1. Purpose

Generate Microsoft Word documents from:

• Pipeline export results (mathpix_pipeline)
• Structured data (JSON/YAML/Dict)
• Template-based reports

2. Prerequisites

pip install python-docx>=0.8.11

3. Invocation

# Generate from pipeline results
/docx-automation generate <source_file> [--output <output.docx>]

# Convert JSON to DOCX
/docx-automation from-json <data.json> --output report.docx

# Generate report from pipeline data
/docx-automation report <image_id> --format detailed

# With workload context
/docx-automation generate synthesis-report --workload <slug>

4. L1/L2/L3 Output Format

L1 Summary (returned to main context)

taskId: docx-{timestamp}
agentType: docx-automation
status: success
summary: "Generated report.docx (125KB)"

filePath: ".agent/prompts/{slug}/docx/report.docx"
fileSize: 128000
pageCount: 5

l2Path: .agent/prompts/{slug}/docx-automation/l2_index.md
requiresL2Read: false
nextActionHint: "Document at {filePath}"

L2 Report Structure

# DOCX Generation Summary

## Document Info
- **Output:** report.docx
- **Size:** 125 KB
- **Pages:** 5

## Content Structure
1. Title Section
2. Content (equations, graphs)
3. Metadata
4. Provenance

## Generation Details
- Font: Times New Roman, 11pt
- Margins: 1 inch
- Header/Footer: Enabled

5. Actions

generate

Generate DOCX from pipeline results.

/docx-automation generate <source_file> [--output <output.docx>] [--config <config.yaml>]

from-json

Convert structured JSON to formatted DOCX.

/docx-automation from-json <data.json> --output <output.docx> [--template <template>]

report

Generate detailed report from pipeline results.

/docx-automation report <image_id> --format <summary|detailed|full> [--include-images]

6. Configuration Options

7. Programmatic Usage

from mathpix_pipeline.export.exporters.docx_exporter import (
    DOCXExporter,
    DOCXExporterConfig,
)
from mathpix_pipeline.schemas.export import ExportOptions

# Initialize with custom config
config = DOCXExporterConfig(
    output_dir=Path("./exports"),
    font_name="Arial",
    font_size_pt=12,
)
exporter = DOCXExporter(config)

# Export
options = ExportOptions(include_metadata=True)
spec = exporter.export(data, options, image_id="img_001")
print(f"Exported to: {spec.file_path}")

8. Output Structure

Generated DOCX documents include:

1. Title Section - Document title, center-aligned
2. Content Section - Outputs, graphs, equations
3. Metadata Section (optional) - Generation timestamp, attributes
4. Provenance Section (optional) - Pipeline version, stages

9. Error Handling

try:
    spec = exporter.export(data, options, image_id)
except ImportError:
    print("Install python-docx: pip install python-docx")
except ExporterError as e:
    print(f"Export error: {e.message}")

10. Pipeline Integration

[Independent Utility]
    |
/docx-automation ---- Can be called standalone
    |
    +-- Output: .agent/prompts/{slug}/docx/{filename}.docx

Workload Context Resolution

# Priority:
# 1. --workload argument
# 2. Active workload (_active_workload.yaml)
# 3. New workload (document type based)

11. Handoff Contract

handoff:
  skill: "docx-automation"
  workload_slug: "{slug}"
  status: "completed"
  next_action:
    skill: null
    arguments: null
    required: false
    reason: "Document generated successfully"
  output:
    path: ".agent/prompts/{slug}/docx/{filename}.docx"
    format: "docx"

12. Post-Compact Recovery

if (isPostCompactSession()) {
  const slug = await getActiveWorkload();
  if (slug) {
    const exportDir = `.agent/prompts/${slug}/docx/`;
    const partialFiles = await Glob(`${exportDir}/*.partial`);
    if (partialFiles.length > 0) {
      console.log("Resuming partial document generation...");
    }
  }
}

Version History