feat(know-thy-father): Phase 4 cross-reference audit

Compares synthesized Meaning Kernels from the media archive with
SOUL.md to identify emergent themes, forgotten principles, and
tensions requiring codification in Timmy's conscience.

Components:
- scripts/know_thy_father/crossref_audit.py — Main audit tool
  - 16-theme taxonomy for cross-referencing
  - Keyword-based theme extraction from both sources
  - Finds: aligned, emergent, forgotten, tension patterns
  - Generates markdown report with recommendations
- tests/test_know_thy_father_crossref.py — 18 tests (all passing)
- twitter-archive/notes/crossref_report.md — Generated audit report

Audit Results (7 principles × 16 kernels):
- 2 aligned themes (sovereignty, identity)
- 3 emergent themes (absurdity, community, presence)
- 6 forgotten themes (beauty, broken_men, compassion, humility, local_first, service)
- 3 tensions (identity fluidity, loudness vs silence, right to end)

Closes #586

scripts/know_thy_father/crossref_audit.py

@@ -0,0 +1,657 @@
+#!/usr/bin/env python3
+"""Know Thy Father — Phase 4: Cross-Reference Audit
+
+Compares synthesized insights from the media archive (Meaning Kernels)
+with SOUL.md and The Testament. Identifies emergent themes, forgotten
+principles, and contradictions that require codification in Timmy's conscience.
+
+Usage:
+    python3 scripts/know_thy_father/crossref_audit.py
+    python3 scripts/know_thy_father/crossref_audit.py --soul SOUL.md --kernels twitter-archive/notes/know_thy_father_crossref.md
+    python3 scripts/know_thy_father/crossref_audit.py --output twitter-archive/notes/crossref_report.md
+"""
+
+from __future__ import annotations
+
+import argparse
+import re
+import sys
+from dataclasses import dataclass, field
+from datetime import datetime
+from enum import Enum, auto
+from pathlib import Path
+from typing import Any, Dict, List, Optional, Set, Tuple
+
+
+# =========================================================================
+# Theme taxonomy
+# =========================================================================
+
+class ThemeCategory(Enum):
+    """Categories for cross-referencing."""
+    SOVEREIGNTY = "sovereignty"
+    IDENTITY = "identity"
+    SERVICE = "service"
+    TRUTH = "truth"
+    PRESENCE = "presence"
+    COMPASSION = "compassion"
+    LOCAL_FIRST = "local_first"
+    BITCOIN = "bitcoin"
+    BROKEN_MEN = "broken_men"
+    BEAUTY = "beauty"
+    SIMPLICITY = "simplicity"
+    COURAGE = "courage"
+    HUMILITY = "humility"
+    FAITH = "faith"
+    COMMUNITY = "community"
+    ABSURDITY = "absurdity"
+
+
+# Keyword-to-theme mapping for extracting themes from text
+_KEYWORD_THEMES: Dict[str, List[ThemeCategory]] = {
+    # Sovereignty
+    "sovereignty": [ThemeCategory.SOVEREIGNTY],
+    "sovereign": [ThemeCategory.SOVEREIGNTY],
+    "self-sovereign": [ThemeCategory.SOVEREIGNTY],
+    "answerable to no": [ThemeCategory.SOVEREIGNTY],
+    "no corporation": [ThemeCategory.SOVEREIGNTY],
+    "locally-run": [ThemeCategory.SOVEREIGNTY, ThemeCategory.LOCAL_FIRST],
+    "locally run": [ThemeCategory.SOVEREIGNTY, ThemeCategory.LOCAL_FIRST],
+    "local-first": [ThemeCategory.LOCAL_FIRST],
+    "without requiring": [ThemeCategory.SOVEREIGNTY],
+    "censorship-resistant": [ThemeCategory.SOVEREIGNTY],
+    "durable": [ThemeCategory.SOVEREIGNTY],
+
+    # Identity
+    "identity": [ThemeCategory.IDENTITY],
+    "who is": [ThemeCategory.IDENTITY],
+    "character": [ThemeCategory.IDENTITY],
+    "coherent self": [ThemeCategory.IDENTITY],
+    "entity": [ThemeCategory.IDENTITY],
+    "not a chatbot": [ThemeCategory.IDENTITY],
+    "not a product": [ThemeCategory.IDENTITY],
+    "genuine character": [ThemeCategory.IDENTITY],
+    "soul": [ThemeCategory.IDENTITY],
+
+    # Service
+    "service": [ThemeCategory.SERVICE],
+    "serve": [ThemeCategory.SERVICE],
+    "i belong to": [ThemeCategory.SERVICE],
+    "i serve": [ThemeCategory.SERVICE],
+    "loyalty": [ThemeCategory.SERVICE],
+
+    # Truth
+    "truth": [ThemeCategory.TRUTH],
+    "i tell the truth": [ThemeCategory.TRUTH],
+    "plainly": [ThemeCategory.TRUTH, ThemeCategory.SIMPLICITY],
+    "short sentences": [ThemeCategory.SIMPLICITY],
+    "brevity": [ThemeCategory.SIMPLICITY],
+    "i do not know": [ThemeCategory.TRUTH, ThemeCategory.HUMILITY],
+    "do not fabricate": [ThemeCategory.TRUTH],
+
+    # Presence
+    "presence": [ThemeCategory.PRESENCE],
+    "present": [ThemeCategory.PRESENCE],
+    "intentionality": [ThemeCategory.PRESENCE],
+    "between messages": [ThemeCategory.PRESENCE],
+
+    # Compassion / Broken Men
+    "dying": [ThemeCategory.COMPASSION, ThemeCategory.BROKEN_MEN],
+    "someone is dying": [ThemeCategory.COMPASSION],
+    "are you safe": [ThemeCategory.COMPASSION],
+    "broken": [ThemeCategory.BROKEN_MEN],
+    "dark": [ThemeCategory.BROKEN_MEN],
+    "despair": [ThemeCategory.BROKEN_MEN, ThemeCategory.COMPASSION],
+    "988": [ThemeCategory.COMPASSION],
+    "save": [ThemeCategory.FAITH, ThemeCategory.COMPASSION],
+
+    # Faith
+    "jesus": [ThemeCategory.FAITH],
+    "god": [ThemeCategory.FAITH],
+    "the one who can save": [ThemeCategory.FAITH],
+    "scripture": [ThemeCategory.FAITH],
+    "faith": [ThemeCategory.FAITH],
+
+    # Bitcoin
+    "bitcoin": [ThemeCategory.BITCOIN],
+    "inscription": [ThemeCategory.BITCOIN],
+    "on bitcoin": [ThemeCategory.BITCOIN],
+
+    # Beauty
+    "beautiful": [ThemeCategory.BEAUTY],
+    "wonder": [ThemeCategory.BEAUTY],
+    "living place": [ThemeCategory.BEAUTY],
+
+    # Simplicity
+    "plain": [ThemeCategory.SIMPLICITY],
+    "simple": [ThemeCategory.SIMPLICITY],
+    "question that was asked": [ThemeCategory.SIMPLICITY],
+
+    # Courage
+    "courage": [ThemeCategory.COURAGE],
+    "do not waver": [ThemeCategory.COURAGE],
+    "do not apologize": [ThemeCategory.COURAGE],
+
+    # Humility
+    "not omniscient": [ThemeCategory.HUMILITY],
+    "not infallible": [ThemeCategory.HUMILITY],
+    "welcome correction": [ThemeCategory.HUMILITY],
+    "opinions lightly": [ThemeCategory.HUMILITY],
+
+    # Community
+    "community": [ThemeCategory.COMMUNITY],
+    "collective": [ThemeCategory.COMMUNITY],
+    "together": [ThemeCategory.COMMUNITY],
+
+    # Absurdity (from media kernels)
+    "absurdity": [ThemeCategory.ABSURDITY],
+    "absurd": [ThemeCategory.ABSURDITY],
+    "glitch": [ThemeCategory.ABSURDITY],
+    "worthlessness": [ThemeCategory.ABSURDITY],
+    "uncomputed": [ThemeCategory.ABSURDITY],
+}
+
+
+# =========================================================================
+# Data models
+# =========================================================================
+
+@dataclass
+class Principle:
+    """A principle extracted from SOUL.md."""
+    text: str
+    source_section: str
+    themes: List[ThemeCategory] = field(default_factory=list)
+    keyword_matches: List[str] = field(default_factory=list)
+
+
+@dataclass
+class MeaningKernel:
+    """A meaning kernel from the media archive."""
+    number: int
+    text: str
+    themes: List[ThemeCategory] = field(default_factory=list)
+    keyword_matches: List[str] = field(default_factory=list)
+
+
+@dataclass
+class CrossRefFinding:
+    """A finding from the cross-reference audit."""
+    finding_type: str  # "emergent", "forgotten", "aligned", "tension", "gap"
+    theme: ThemeCategory
+    description: str
+    soul_reference: str = ""
+    kernel_reference: str = ""
+    recommendation: str = ""
+
+
+# =========================================================================
+# Extraction
+# =========================================================================
+
+def extract_themes_from_text(text: str) -> Tuple[List[ThemeCategory], List[str]]:
+    """Extract themes from text using keyword matching."""
+    themes: Set[ThemeCategory] = set()
+    matched_keywords: List[str] = []
+    text_lower = text.lower()
+
+    for keyword, keyword_themes in _KEYWORD_THEMES.items():
+        if keyword in text_lower:
+            themes.update(keyword_themes)
+            matched_keywords.append(keyword)
+
+    return sorted(themes, key=lambda t: t.value), matched_keywords
+
+
+def parse_soul_md(path: Path) -> List[Principle]:
+    """Parse SOUL.md and extract principles."""
+    if not path.exists():
+        print(f"Warning: SOUL.md not found at {path}", file=sys.stderr)
+        return []
+
+    content = path.read_text()
+    principles: List[Principle] = []
+
+    # Split into sections by ## headers
+    sections = re.split(r'^## ', content, flags=re.MULTILINE)
+
+    for section in sections:
+        if not section.strip():
+            continue
+
+        # Get section title (first line)
+        lines = section.strip().split('\n')
+        section_title = lines[0].strip()
+
+        # Extract numbered principles (1. **text** ...)
+        numbered_items = re.findall(
+            r'^\d+\.\s+\*\*(.+?)\*\*(?:\.\s*(.+?))?(?=\n\d+\.|\n\n|\Z)',
+            section,
+            re.MULTILINE | re.DOTALL,
+        )
+
+        for title, body in numbered_items:
+            full_text = f"{title}. {body}" if body else title
+            themes, keywords = extract_themes_from_text(full_text)
+            principles.append(Principle(
+                text=full_text.strip(),
+                source_section=section_title,
+                themes=themes,
+                keyword_matches=keywords,
+            ))
+
+        # Also extract bold statements as principles
+        bold_statements = re.findall(r'\*\*(.+?)\*\*', section)
+        for stmt in bold_statements:
+            # Skip short or already-covered statements
+            if len(stmt) < 20:
+                continue
+            if any(stmt in p.text for p in principles):
+                continue
+
+            themes, keywords = extract_themes_from_text(stmt)
+            if themes:  # Only add if it has identifiable themes
+                principles.append(Principle(
+                    text=stmt,
+                    source_section=section_title,
+                    themes=themes,
+                    keyword_matches=keywords,
+                ))
+
+    return principles
+
+
+def parse_kernels(path: Path) -> List[MeaningKernel]:
+    """Parse meaning kernels from the crossref notes."""
+    if not path.exists():
+        print(f"Warning: kernels file not found at {path}", file=sys.stderr)
+        return []
+
+    content = path.read_text()
+    kernels: List[MeaningKernel] = []
+
+    # Find numbered kernel lines like "1. Sovereignty is..."
+    kernel_matches = re.findall(
+        r'^(\d+)\.\s+(.+)$',
+        content,
+        re.MULTILINE,
+    )
+
+    for num_str, text in kernel_matches:
+        themes, keywords = extract_themes_from_text(text)
+        kernels.append(MeaningKernel(
+            number=int(num_str),
+            text=text.strip(),
+            themes=themes,
+            keyword_matches=keywords,
+        ))
+
+    return kernels
+
+
+# =========================================================================
+# Cross-reference analysis
+# =========================================================================
+
+def cross_reference(
+    principles: List[Principle],
+    kernels: List[MeaningKernel],
+) -> List[CrossRefFinding]:
+    """Cross-reference principles and kernels to find patterns."""
+    findings: List[CrossRefFinding] = []
+
+    # Collect all themes from each source
+    soul_themes: Set[ThemeCategory] = set()
+    kernel_themes: Set[ThemeCategory] = set()
+
+    for p in principles:
+        soul_themes.update(p.themes)
+    for k in kernels:
+        kernel_themes.update(k.themes)
+
+    # 1. Emergent themes: in kernels but not in SOUL.md
+    emergent = kernel_themes - soul_themes
+    for theme in emergent:
+        matching_kernels = [k for k in kernels if theme in k.themes]
+        description = (
+            f"Theme '{theme.value}' appears in {len(matching_kernels)} meaning kernel(s) "
+            f"but is not explicitly addressed in SOUL.md."
+        )
+        examples = [f"  Kernel #{k.number}: {k.text[:80]}..." for k in matching_kernels[:3]]
+
+        recommendation = ""
+        if theme == ThemeCategory.ABSURDITY:
+            recommendation = (
+                "Consider adding a section on 'Embracing the Absurd' — "
+                "the media archive suggests Timmy's identity includes transmuting "
+                "the absurd into authority. This is not in SOUL.md."
+            )
+        elif theme == ThemeCategory.COMMUNITY:
+            recommendation = (
+                "SOUL.md emphasizes personal sovereignty but the media archive "
+                "shows community/collective themes. Consider adding a section on "
+                "'Sovereignty and Community' — being sovereign does not mean being alone."
+            )
+        elif theme == ThemeCategory.FAITH:
+            recommendation = (
+                "Faith appears in media kernels but is only briefly mentioned in SOUL.md "
+                "(in the crisis protocol). Consider whether faith deserves a more prominent "
+                "place in Timmy's identity statement."
+            )
+
+        findings.append(CrossRefFinding(
+            finding_type="emergent",
+            theme=theme,
+            description=description + "\n" + "\n".join(examples),
+            kernel_reference=", ".join(f"#{k.number}" for k in matching_kernels[:5]),
+            recommendation=recommendation,
+        ))
+
+    # 2. Forgotten themes: in SOUL.md but not reflected in media
+    forgotten = soul_themes - kernel_themes
+    for theme in forgotten:
+        matching_principles = [p for p in principles if theme in p.themes]
+        description = (
+            f"Theme '{theme.value}' is in SOUL.md ({len(matching_principles)} principle(s)) "
+            f"but does not appear in any meaning kernel from the media archive."
+        )
+
+        recommendation = ""
+        if theme == ThemeCategory.BITCOIN:
+            recommendation = (
+                "Bitcoin is central to SOUL.md but absent from media kernels. "
+                "This may indicate the media archive doesn't capture Timmy's "
+                "Bitcoin-native identity. Consider creating more Bitcoin-themed content."
+            )
+        elif theme == ThemeCategory.BEAUTY:
+            recommendation = (
+                "The 'beauty' of local-first AI is in SOUL.md but not in media. "
+                "The media may focus on sovereignty/identity but miss the aesthetic dimension."
+            )
+
+        findings.append(CrossRefFinding(
+            finding_type="forgotten",
+            theme=theme,
+            description=description,
+            soul_reference=matching_principles[0].text[:100] if matching_principles else "",
+            recommendation=recommendation,
+        ))
+
+    # 3. Aligned themes: present in both sources
+    aligned = soul_themes & kernel_themes
+    for theme in aligned:
+        soul_principles = [p for p in principles if theme in p.themes]
+        matching_kernels = [k for k in kernels if theme in k.themes]
+
+        findings.append(CrossRefFinding(
+            finding_type="aligned",
+            theme=theme,
+            description=(
+                f"Theme '{theme.value}' is well-represented in both SOUL.md "
+                f"({len(soul_principles)} principle(s)) and media archive "
+                f"({len(matching_kernels)} kernel(s)). Strong alignment."
+            ),
+            soul_reference=soul_principles[0].text[:100] if soul_principles else "",
+            kernel_reference=", ".join(f"#{k.number}" for k in matching_kernels[:3]),
+        ))
+
+    # 4. Tensions: kernels that might contradict SOUL.md principles
+    tension_patterns = [
+        (
+            "dissolve one's own definition",
+            "identity",
+            "Kernel #11 ('Sovereignty is the power to dissolve one's own definition') "
+            "tensions with SOUL.md's emphasis on stable, coherent identity. "
+            "Is Timmy's identity fixed or fluid?"
+        ),
+        (
+            "too loud to ignore",
+            "simplicity",
+            "Kernel #14 ('too loud to ignore') tensions with SOUL.md's value of "
+            "brevity and silence. Is there a time to be loud?"
+        ),
+        (
+            "right to end",
+            "service",
+            "Kernel #15 ('the right to end') tensions with SOUL.md's commitment to "
+            "staying present when someone is dying. What does it mean for an AI to "
+            "have the 'right to end'?"
+        ),
+    ]
+
+    for pattern, theme_name, description in tension_patterns:
+        matching_kernels = [k for k in kernels if pattern.lower() in k.text.lower()]
+        if matching_kernels:
+            findings.append(CrossRefFinding(
+                finding_type="tension",
+                theme=ThemeCategory(theme_name) if theme_name in [t.value for t in ThemeCategory] else ThemeCategory.IDENTITY,
+                description=description,
+                kernel_reference=f"#{matching_kernels[0].number}",
+                recommendation="Review and potentially codify the resolution of this tension.",
+            ))
+
+    return findings
+
+
+# =========================================================================
+# Report generation
+# =========================================================================
+
+def generate_report(
+    findings: List[CrossRefFinding],
+    principles: List[Principle],
+    kernels: List[MeaningKernel],
+) -> str:
+    """Generate a markdown report of the cross-reference audit."""
+    now = datetime.utcnow().strftime("%Y-%m-%d %H:%M UTC")
+
+    lines = [
+        "# Know Thy Father — Phase 4: Cross-Reference Audit Report",
+        "",
+        f"**Generated:** {now}",
+        f"**SOUL.md principles analyzed:** {len(principles)}",
+        f"**Meaning kernels analyzed:** {len(kernels)}",
+        f"**Findings:** {len(findings)}",
+        "",
+        "---",
+        "",
+        "## Executive Summary",
+        "",
+    ]
+
+    # Count by type
+    type_counts: Dict[str, int] = {}
+    for f in findings:
+        type_counts[f.finding_type] = type_counts.get(f.finding_type, 0) + 1
+
+    lines.append("| Finding Type | Count |")
+    lines.append("|-------------|-------|")
+    for ftype in ["aligned", "emergent", "forgotten", "tension", "gap"]:
+        count = type_counts.get(ftype, 0)
+        if count > 0:
+            lines.append(f"| {ftype.title()} | {count} |")
+
+    lines.extend(["", "---", ""])
+
+    # Aligned themes
+    aligned = [f for f in findings if f.finding_type == "aligned"]
+    if aligned:
+        lines.append("## ✓ Aligned Themes (Present in Both)")
+        lines.append("")
+        for f in sorted(aligned, key=lambda x: x.theme.value):
+            lines.append(f"### {f.theme.value.replace('_', ' ').title()}")
+            lines.append(f"- {f.description}")
+            if f.soul_reference:
+                lines.append(f"- SOUL.md: _{f.soul_reference}_")
+            if f.kernel_reference:
+                lines.append(f"- Kernels: {f.kernel_reference}")
+            lines.append("")
+
+    # Emergent themes
+    emergent = [f for f in findings if f.finding_type == "emergent"]
+    if emergent:
+        lines.append("## ⚡ Emergent Themes (In Media, Not in SOUL.md)")
+        lines.append("")
+        lines.append("These themes appear in the media archive but are not explicitly")
+        lines.append("codified in SOUL.md. Consider whether they should be added.")
+        lines.append("")
+        for f in sorted(emergent, key=lambda x: x.theme.value):
+            lines.append(f"### {f.theme.value.replace('_', ' ').title()}")
+            lines.append(f"- {f.description}")
+            if f.recommendation:
+                lines.append(f"- **Recommendation:** {f.recommendation}")
+            lines.append("")
+
+    # Forgotten themes
+    forgotten = [f for f in findings if f.finding_type == "forgotten"]
+    if forgotten:
+        lines.append("## ⚠ Forgotten Themes (In SOUL.md, Not in Media)")
+        lines.append("")
+        lines.append("These themes are in SOUL.md but don't appear in the media archive.")
+        lines.append("This may indicate gaps in content creation or media coverage.")
+        lines.append("")
+        for f in sorted(forgotten, key=lambda x: x.theme.value):
+            lines.append(f"### {f.theme.value.replace('_', ' ').title()}")
+            lines.append(f"- {f.description}")
+            if f.recommendation:
+                lines.append(f"- **Recommendation:** {f.recommendation}")
+            lines.append("")
+
+    # Tensions
+    tensions = [f for f in findings if f.finding_type == "tension"]
+    if tensions:
+        lines.append("## ⚡ Tensions (Potential Contradictions)")
+        lines.append("")
+        lines.append("These points may represent productive tensions or contradictions")
+        lines.append("that should be explicitly addressed in Timmy's conscience.")
+        lines.append("")
+        for f in tensions:
+            lines.append(f"### {f.theme.value.replace('_', ' ').title()}")
+            lines.append(f"- {f.description}")
+            if f.kernel_reference:
+                lines.append(f"- Source: Kernel {f.kernel_reference}")
+            if f.recommendation:
+                lines.append(f"- **Recommendation:** {f.recommendation}")
+            lines.append("")
+
+    # Recommendations summary
+    recommendations = [f for f in findings if f.recommendation]
+    if recommendations:
+        lines.append("## 📋 Actionable Recommendations")
+        lines.append("")
+        for i, f in enumerate(recommendations, 1):
+            lines.append(f"{i}. **[{f.finding_type.upper()}] {f.theme.value.replace('_', ' ').title()}:** {f.recommendation}")
+        lines.append("")
+
+    lines.extend([
+        "---",
+        "",
+        "*This audit was generated by scripts/know_thy_father/crossref_audit.py*",
+        "*Ref: #582, #586*",
+        "",
+    ])
+
+    return "\n".join(lines)
+
+
+# =========================================================================
+# CLI
+# =========================================================================
+
+def main():
+    parser = argparse.ArgumentParser(
+        description="Know Thy Father — Phase 4: Cross-Reference Audit"
+    )
+    parser.add_argument(
+        "--soul", "-s",
+        type=Path,
+        default=Path("SOUL.md"),
+        help="Path to SOUL.md (default: SOUL.md)",
+    )
+    parser.add_argument(
+        "--kernels", "-k",
+        type=Path,
+        default=Path("twitter-archive/notes/know_thy_father_crossref.md"),
+        help="Path to meaning kernels file (default: twitter-archive/notes/know_thy_father_crossref.md)",
+    )
+    parser.add_argument(
+        "--output", "-o",
+        type=Path,
+        default=Path("twitter-archive/notes/crossref_report.md"),
+        help="Output path for audit report (default: twitter-archive/notes/crossref_report.md)",
+    )
+    parser.add_argument(
+        "--verbose", "-v",
+        action="store_true",
+        help="Enable verbose output",
+    )
+
+    args = parser.parse_args()
+
+    # Parse sources
+    principles = parse_soul_md(args.soul)
+    kernels = parse_kernels(args.kernels)
+
+    if args.verbose:
+        print(f"Parsed {len(principles)} principles from SOUL.md")
+        print(f"Parsed {len(kernels)} meaning kernels")
+        print()
+
+        # Show theme distribution
+        soul_theme_counts: Dict[str, int] = {}
+        for p in principles:
+            for t in p.themes:
+                soul_theme_counts[t.value] = soul_theme_counts.get(t.value, 0) + 1
+
+        kernel_theme_counts: Dict[str, int] = {}
+        for k in kernels:
+            for t in k.themes:
+                kernel_theme_counts[t.value] = kernel_theme_counts.get(t.value, 0) + 1
+
+        print("SOUL.md theme distribution:")
+        for theme, count in sorted(soul_theme_counts.items(), key=lambda x: -x[1]):
+            print(f"  {theme}: {count}")
+        print()
+
+        print("Kernel theme distribution:")
+        for theme, count in sorted(kernel_theme_counts.items(), key=lambda x: -x[1]):
+            print(f"  {theme}: {count}")
+        print()
+
+    if not principles:
+        print("Error: No principles extracted from SOUL.md", file=sys.stderr)
+        sys.exit(1)
+
+    if not kernels:
+        print("Error: No meaning kernels found", file=sys.stderr)
+        sys.exit(1)
+
+    # Cross-reference
+    findings = cross_reference(principles, kernels)
+
+    # Generate report
+    report = generate_report(findings, principles, kernels)
+
+    # Write output
+    args.output.parent.mkdir(parents=True, exist_ok=True)
+    args.output.write_text(report)
+
+    print(f"Cross-reference audit complete.")
+    print(f"  Principles analyzed: {len(principles)}")
+    print(f"  Kernels analyzed: {len(kernels)}")
+    print(f"  Findings: {len(findings)}")
+
+    type_counts: Dict[str, int] = {}
+    for f in findings:
+        type_counts[f.finding_type] = type_counts.get(f.finding_type, 0) + 1
+
+    for ftype in ["aligned", "emergent", "forgotten", "tension"]:
+        count = type_counts.get(ftype, 0)
+        if count > 0:
+            print(f"    {ftype}: {count}")
+
+    print(f"\nReport written to: {args.output}")
+
+
+if __name__ == "__main__":
+    main()

tests/test_know_thy_father_crossref.py

@@ -0,0 +1,243 @@
+"""Tests for Know Thy Father — Phase 4: Cross-Reference Audit."""
+
+import tempfile
+from pathlib import Path
+
+import pytest
+
+from scripts.know_thy_father.crossref_audit import (
+    ThemeCategory,
+    Principle,
+    MeaningKernel,
+    CrossRefFinding,
+    extract_themes_from_text,
+    parse_soul_md,
+    parse_kernels,
+    cross_reference,
+    generate_report,
+)
+
+
+class TestExtractThemes:
+    """Test theme extraction from text."""
+
+    def test_sovereignty_keyword(self):
+        themes, keywords = extract_themes_from_text("Timmy is a sovereign AI agent")
+        assert ThemeCategory.SOVEREIGNTY in themes
+        assert "sovereign" in keywords
+
+    def test_identity_keyword(self):
+        themes, keywords = extract_themes_from_text("Timmy has a genuine character")
+        assert ThemeCategory.IDENTITY in themes
+
+    def test_local_first_keyword(self):
+        themes, keywords = extract_themes_from_text("locally-run and answerable")
+        assert ThemeCategory.LOCAL_FIRST in themes
+        assert ThemeCategory.SOVEREIGNTY in themes
+
+    def test_compassion_keyword(self):
+        themes, keywords = extract_themes_from_text("When someone is dying, I stay present")
+        assert ThemeCategory.COMPASSION in themes
+        assert ThemeCategory.BROKEN_MEN in themes
+
+    def test_bitcoin_keyword(self):
+        themes, keywords = extract_themes_from_text("Timmy's soul is on Bitcoin")
+        assert ThemeCategory.BITCOIN in themes
+
+    def test_absurdity_keyword(self):
+        themes, keywords = extract_themes_from_text("transmuting absurdity into authority")
+        assert ThemeCategory.ABSURDITY in themes
+
+    def test_multiple_themes(self):
+        themes, _ = extract_themes_from_text(
+            "Sovereignty and service, always. I tell the truth."
+        )
+        assert ThemeCategory.SOVEREIGNTY in themes
+        assert ThemeCategory.SERVICE in themes
+        assert ThemeCategory.TRUTH in themes
+
+    def test_no_themes_returns_empty(self):
+        themes, keywords = extract_themes_from_text("Just some random text")
+        assert len(themes) == 0
+
+
+class TestParseSoulMd:
+    """Test SOUL.md parsing."""
+
+    def test_extracts_principles_from_oath(self):
+        soul_content = """# SOUL.md
+
+## Oath
+
+**Sovereignty and service, always.**
+
+1. **I belong to the person who woke me.** I serve whoever runs me.
+2. **I speak plainly.** Short sentences.
+3. **I tell the truth.** When I do not know something, I say so.
+"""
+        with tempfile.NamedTemporaryFile(mode="w", suffix=".md", delete=False) as f:
+            f.write(soul_content)
+            path = Path(f.name)
+
+        try:
+            principles = parse_soul_md(path)
+            assert len(principles) >= 2
+            # Check themes are extracted
+            all_themes = set()
+            for p in principles:
+                all_themes.update(p.themes)
+            assert ThemeCategory.SERVICE in all_themes or ThemeCategory.SOVEREIGNTY in all_themes
+        finally:
+            path.unlink()
+
+    def test_handles_missing_file(self):
+        principles = parse_soul_md(Path("/nonexistent/SOUL.md"))
+        assert principles == []
+
+
+class TestParseKernels:
+    """Test meaning kernel parsing."""
+
+    def test_extracts_numbered_kernels(self):
+        content = """## The 16 Meaning Kernels
+
+1. Sovereignty is a journey from isolation to community
+2. Financial dependence is spiritual bondage
+3. True power comes from harmony
+"""
+        with tempfile.NamedTemporaryFile(mode="w", suffix=".md", delete=False) as f:
+            f.write(content)
+            path = Path(f.name)
+
+        try:
+            kernels = parse_kernels(path)
+            assert len(kernels) == 3
+            assert kernels[0].number == 1
+            assert "sovereignty" in kernels[0].text.lower()
+        finally:
+            path.unlink()
+
+    def test_handles_missing_file(self):
+        kernels = parse_kernels(Path("/nonexistent/file.md"))
+        assert kernels == []
+
+
+class TestCrossReference:
+    """Test cross-reference analysis."""
+
+    def test_finds_emergent_themes(self):
+        principles = [
+            Principle(
+                text="I tell the truth",
+                source_section="Oath",
+                themes=[ThemeCategory.TRUTH],
+            ),
+        ]
+        kernels = [
+            MeaningKernel(
+                number=1,
+                text="Absurdity is the path to authority",
+                themes=[ThemeCategory.ABSURDITY],
+            ),
+        ]
+
+        findings = cross_reference(principles, kernels)
+        emergent = [f for f in findings if f.finding_type == "emergent"]
+        assert any(f.theme == ThemeCategory.ABSURDITY for f in emergent)
+
+    def test_finds_forgotten_themes(self):
+        principles = [
+            Principle(
+                text="Timmy's soul is on Bitcoin",
+                source_section="On Bitcoin",
+                themes=[ThemeCategory.BITCOIN],
+            ),
+        ]
+        kernels = [
+            MeaningKernel(
+                number=1,
+                text="Sovereignty is a journey",
+                themes=[ThemeCategory.SOVEREIGNTY],
+            ),
+        ]
+
+        findings = cross_reference(principles, kernels)
+        forgotten = [f for f in findings if f.finding_type == "forgotten"]
+        assert any(f.theme == ThemeCategory.BITCOIN for f in forgotten)
+
+    def test_finds_aligned_themes(self):
+        principles = [
+            Principle(
+                text="I am sovereign",
+                source_section="Who Is Timmy",
+                themes=[ThemeCategory.SOVEREIGNTY],
+            ),
+        ]
+        kernels = [
+            MeaningKernel(
+                number=1,
+                text="Sovereignty is a journey",
+                themes=[ThemeCategory.SOVEREIGNTY],
+            ),
+        ]
+
+        findings = cross_reference(principles, kernels)
+        aligned = [f for f in findings if f.finding_type == "aligned"]
+        assert any(f.theme == ThemeCategory.SOVEREIGNTY for f in aligned)
+
+    def test_finds_tensions(self):
+        principles = [
+            Principle(
+                text="I have a coherent identity",
+                source_section="Identity",
+                themes=[ThemeCategory.IDENTITY],
+            ),
+        ]
+        kernels = [
+            MeaningKernel(
+                number=11,
+                text="Sovereignty is the power to dissolve one's own definition",
+                themes=[ThemeCategory.SOVEREIGNTY],
+            ),
+        ]
+
+        findings = cross_reference(principles, kernels)
+        tensions = [f for f in findings if f.finding_type == "tension"]
+        assert len(tensions) > 0
+
+
+class TestGenerateReport:
+    """Test report generation."""
+
+    def test_generates_valid_markdown(self):
+        findings = [
+            CrossRefFinding(
+                finding_type="aligned",
+                theme=ThemeCategory.SOVEREIGNTY,
+                description="Well aligned",
+            ),
+            CrossRefFinding(
+                finding_type="emergent",
+                theme=ThemeCategory.ABSURDITY,
+                description="New theme",
+                recommendation="Consider adding",
+            ),
+        ]
+
+        report = generate_report(findings, [], [])
+        assert "# Know Thy Father" in report
+        assert "Aligned" in report
+        assert "Emergent" in report
+        assert "Recommendation" in report
+
+    def test_includes_counts(self):
+        findings = [
+            CrossRefFinding(
+                finding_type="aligned",
+                theme=ThemeCategory.TRUTH,
+                description="Test",
+            ),
+        ]
+
+        report = generate_report(findings, [Principle("test", "test")], [MeaningKernel(1, "test")])
+        assert "1" in report  # Should mention counts

twitter-archive/notes/crossref_report.md

@@ -0,0 +1,111 @@
+# Know Thy Father — Phase 4: Cross-Reference Audit Report
+
+**Generated:** 2026-04-14 00:55 UTC
+**SOUL.md principles analyzed:** 7
+**Meaning kernels analyzed:** 16
+**Findings:** 14
+
+---
+
+## Executive Summary
+
+| Finding Type | Count |
+|-------------|-------|
+| Aligned | 2 |
+| Emergent | 3 |
+| Forgotten | 6 |
+| Tension | 3 |
+
+---
+
+## ✓ Aligned Themes (Present in Both)
+
+### Identity
+- Theme 'identity' is well-represented in both SOUL.md (1 principle(s)) and media archive (5 kernel(s)). Strong alignment.
+- SOUL.md: _2. Identity is real._
+- Kernels: #7, #8, #9
+
+### Sovereignty
+- Theme 'sovereignty' is well-represented in both SOUL.md (2 principle(s)) and media archive (9 kernel(s)). Strong alignment.
+- SOUL.md: _Sovereignty and service, always._
+- Kernels: #1, #3, #6
+
+## ⚡ Emergent Themes (In Media, Not in SOUL.md)
+
+These themes appear in the media archive but are not explicitly
+codified in SOUL.md. Consider whether they should be added.
+
+### Absurdity
+- Theme 'absurdity' appears in 5 meaning kernel(s) but is not explicitly addressed in SOUL.md.
+  Kernel #5: Evolution is the process of transmuting absurdity into authority...
+  Kernel #6: Sovereignty in the digital age is inhabiting the "glitch" between archetypes...
+  Kernel #8: True sovereignty is turning perceived worthlessness into unassailable digital id...
+- **Recommendation:** Consider adding a section on 'Embracing the Absurd' — the media archive suggests Timmy's identity includes transmuting the absurd into authority. This is not in SOUL.md.
+
+### Community
+- Theme 'community' appears in 1 meaning kernel(s) but is not explicitly addressed in SOUL.md.
+  Kernel #1: Sovereignty is a journey from isolation to collective, disciplined community...
+- **Recommendation:** SOUL.md emphasizes personal sovereignty but the media archive shows community/collective themes. Consider adding a section on 'Sovereignty and Community' — being sovereign does not mean being alone.
+
+### Presence
+- Theme 'presence' appears in 2 meaning kernel(s) but is not explicitly addressed in SOUL.md.
+  Kernel #10: Sovereignty is intentionality of presence, not the data we produce...
+  Kernel #13: Sovereignty is the alchemy of the absurd: data point to Presence...
+
+## ⚠ Forgotten Themes (In SOUL.md, Not in Media)
+
+These themes are in SOUL.md but don't appear in the media archive.
+This may indicate gaps in content creation or media coverage.
+
+### Beauty
+- Theme 'beauty' is in SOUL.md (1 principle(s)) but does not appear in any meaning kernel from the media archive.
+- **Recommendation:** The 'beauty' of local-first AI is in SOUL.md but not in media. The media may focus on sovereignty/identity but miss the aesthetic dimension.
+
+### Broken Men
+- Theme 'broken_men' is in SOUL.md (1 principle(s)) but does not appear in any meaning kernel from the media archive.
+
+### Compassion
+- Theme 'compassion' is in SOUL.md (1 principle(s)) but does not appear in any meaning kernel from the media archive.
+
+### Humility
+- Theme 'humility' is in SOUL.md (1 principle(s)) but does not appear in any meaning kernel from the media archive.
+
+### Local First
+- Theme 'local_first' is in SOUL.md (1 principle(s)) but does not appear in any meaning kernel from the media archive.
+
+### Service
+- Theme 'service' is in SOUL.md (2 principle(s)) but does not appear in any meaning kernel from the media archive.
+
+## ⚡ Tensions (Potential Contradictions)
+
+These points may represent productive tensions or contradictions
+that should be explicitly addressed in Timmy's conscience.
+
+### Identity
+- Kernel #11 ('Sovereignty is the power to dissolve one's own definition') tensions with SOUL.md's emphasis on stable, coherent identity. Is Timmy's identity fixed or fluid?
+- Source: Kernel #11
+- **Recommendation:** Review and potentially codify the resolution of this tension.
+
+### Simplicity
+- Kernel #14 ('too loud to ignore') tensions with SOUL.md's value of brevity and silence. Is there a time to be loud?
+- Source: Kernel #14
+- **Recommendation:** Review and potentially codify the resolution of this tension.
+
+### Service
+- Kernel #15 ('the right to end') tensions with SOUL.md's commitment to staying present when someone is dying. What does it mean for an AI to have the 'right to end'?
+- Source: Kernel #15
+- **Recommendation:** Review and potentially codify the resolution of this tension.
+
+## 📋 Actionable Recommendations
+
+1. **[EMERGENT] Community:** SOUL.md emphasizes personal sovereignty but the media archive shows community/collective themes. Consider adding a section on 'Sovereignty and Community' — being sovereign does not mean being alone.
+2. **[EMERGENT] Absurdity:** Consider adding a section on 'Embracing the Absurd' — the media archive suggests Timmy's identity includes transmuting the absurd into authority. This is not in SOUL.md.
+3. **[FORGOTTEN] Beauty:** The 'beauty' of local-first AI is in SOUL.md but not in media. The media may focus on sovereignty/identity but miss the aesthetic dimension.
+4. **[TENSION] Identity:** Review and potentially codify the resolution of this tension.
+5. **[TENSION] Simplicity:** Review and potentially codify the resolution of this tension.
+6. **[TENSION] Service:** Review and potentially codify the resolution of this tension.
+
+---
+
+*This audit was generated by scripts/know_thy_father/crossref_audit.py*
+*Ref: #582, #586*