Changelog - 2025-12-06 (#2)

Scorecard Integration into Workflow

Overview

This release adds automatic 12Ps scorecard integration to the memo workflow. Previously, the scorecard evaluator generated detailed scores in 5-scorecard/12Ps-scorecard.md, but this ran AFTER the writer created section 8—meaning the scorecard data never made it into the final draft. Now, a new integrate_scorecard workflow step automatically merges the scorecard into section 8 and reassembles the final draft.

Problem

The workflow sequence had a timing issue:

writer (creates section 8 "Scorecard Summary") → ... → scorecard_evaluator (generates 5-scorecard/12Ps-scorecard.md)

Issues:

1. Section 8 was written as a placeholder without actual scorecard data

2. The detailed 12Ps evaluation was generated but stored separately in 5-scorecard/

3. Users had to manually copy the scorecard into the final document

4. The final draft shipped without the core investment evaluation

Solution: integrate_scorecard Workflow Node

Added a new workflow node that runs after scorecard evaluation:

scorecard → integrate_scorecard → conditional(should_continue) → finalize/human_review

What integrate_scorecard() Does

1. Locates scorecard file at {output_dir}/5-scorecard/12Ps-scorecard.md

2. Finds section 8 by pattern {output_dir}/2-sections/08-*.md or *scorecard*.md

3. Replaces section 8 content with the full scorecard

4. Reassembles final draft from all sections in order

5. Preserves header (logo/trademark) if present from existing draft

Function Implementation

def integrate_scorecard(state: MemoState) -> dict:
    """
    Integrate the 12Ps scorecard into section 8 and reassemble final draft.
    """
    from .utils import get_latest_output_dir

    company_name = state["company_name"]
    firm = state.get("firm")
    output_dir = get_latest_output_dir(company_name, firm=firm)

    # Check for scorecard file
    scorecard_file = output_dir / "5-scorecard" / "12Ps-scorecard.md"
    if not scorecard_file.exists():
        return {"messages": ["Scorecard integration skipped: no scorecard file"]}

    # Find and replace section 8
    sections_dir = output_dir / "2-sections"
    for f in sections_dir.glob("08-*.md"):
        scorecard_content = scorecard_file.read_text()
        f.write_text(scorecard_content)
        print(f"  ✓ Integrated scorecard into {f.name}")
        break

    # Reassemble final draft preserving header
    final_draft_path = output_dir / "4-final-draft.md"
    content = ""

    # Preserve logo if present
    if final_draft_path.exists():
        existing = final_draft_path.read_text()
        first_line = existing.split('\n')[0]
        if first_line.startswith('!['):
            content = first_line + '\n\n'

    # Assemble all sections
    for section_file in sorted(sections_dir.glob("*.md")):
        if section_file.name == "header.md":
            continue
        content += section_file.read_text() + '\n\n'

    final_draft_path.write_text(content)
    return {"messages": [f"Scorecard integrated ({len(content.split())} words)"]}

Workflow Changes

Node Addition

# In build_workflow()
workflow.add_node("integrate_scorecard", integrate_scorecard)

Edge Updates

Before:

workflow.add_edge("validate", "scorecard")
workflow.add_conditional_edges("scorecard", should_continue, {...})

After:

workflow.add_edge("validate", "scorecard")
workflow.add_edge("scorecard", "integrate_scorecard")
workflow.add_conditional_edges("integrate_scorecard", should_continue, {...})

Updated Flow Sequence

Deck Analyst → Research → Section Research → Draft → Trademark → Socials → Links
→ Visualizations → Citations → TOC → Revise Summaries → Citation Validator
→ Fact Check → Validate → Scorecard → Integrate Scorecard → Finalize/Human Review

Files Changed

Impact

Before This Change

• Final draft section 8 contained placeholder text without actual scores

• Detailed scorecard lived in separate file 5-scorecard/12Ps-scorecard.md

• Users had to manually copy scorecard content into final draft

• Export/review pipeline shipped incomplete memos

After This Change

• Section 8 contains the full 12Ps evaluation with all dimension scores

• Final draft is automatically reassembled with integrated scorecard

• Zero manual intervention required for complete memos

• Scorecard data flows directly into human review/finalization

Example Output

  ✓ Integrated scorecard into 08-12ps-scorecard-summary.md
  ✓ Reassembled final draft: 4,832 words

Testing

Verification Command

# Verify workflow compiles
python -c "from src.workflow import build_workflow; w = build_workflow(); print('✓ Workflow compiles')"

Expected Behavior

After running memo generation:

1. 5-scorecard/12Ps-scorecard.md contains detailed evaluation

2. 2-sections/08-*.md contains same content as scorecard file

3. 4-final-draft.md includes full scorecard in section 8

Related Documentation

• context-vigilance/12Ps-Framework-for-Due-Diligence.md - 12Ps framework specification

• changelog/2025-12-06_01.md - Dimension-grouped questions (same session)

• templates/scorecards/ - Scorecard template definitions

Contributors

Implementation by Claude Code (Anthropic) with user guidance.