feat: add codebase genome diff script for structural change detection

Introduces genome_diff.py — a tool for detecting structural changes between
two git refs: file-level changes, function/class signature modifications,
and dependency import changes.

Addresses #132.

scripts/entity_extractor.py

@@ -1,268 +0,0 @@
-#!/usr/bin/env python3
-"""
-entity_extractor.py — Extract named entities from text sources.
-
-Extracts: people, projects, tools, concepts, repos from session transcripts,
-README files, issue bodies, or any text input.
-
-Output: knowledge/entities.json with deduplicated entity list and occurrence counts.
-"""
-
-import argparse
-import json
-import os
-import sys
-import time
-from datetime import datetime, timezone
-from pathlib import Path
-from typing import Optional
-
-SCRIPT_DIR = Path(__file__).parent.absolute()
-sys.path.insert(0, str(SCRIPT_DIR))
-
-from session_reader import read_session, messages_to_text
-
-# --- Configuration ---
-DEFAULT_API_BASE = os.environ.get("HARVESTER_API_BASE", "https://api.nousresearch.com/v1")
-DEFAULT_API_KEY = os.environ.get("HARVESTER_API_KEY", "")
-DEFAULT_MODEL = os.environ.get("HARVESTER_MODEL", "xiaomi/mimo-v2-pro")
-KNOWLEDGE_DIR = os.environ.get("HARVESTER_KNOWLEDGE_DIR", "knowledge")
-PROMPT_PATH = os.environ.get("ENTITY_PROMPT_PATH", str(SCRIPT_DIR.parent / "templates" / "entity-extraction-prompt.md"))
-
-API_KEY_PATHS = [
-    os.path.expanduser("~/.config/nous/key"),
-    os.path.expanduser("~/.hermes/keymaxxing/active/minimax.key"),
-    os.path.expanduser("~/.config/openrouter/key"),
-]
-
-def find_api_key() -> str:
-    for path in API_KEY_PATHS:
-        if os.path.exists(path):
-            with open(path) as f:
-                key = f.read().strip()
-                if key:
-                    return key
-    return ""
-
-def load_prompt() -> str:
-    path = Path(PROMPT_PATH)
-    if not path.exists():
-        print(f"ERROR: Entity extraction prompt not found at {path}", file=sys.stderr)
-        sys.exit(1)
-    return path.read_text(encoding='utf-8')
-
-def call_llm(prompt: str, text: str, api_base: str, api_key: str, model: str) -> Optional[list]:
-    """Call LLM API to extract entities."""
-    import urllib.request
-    
-    messages = [
-        {"role": "system", "content": prompt},
-        {"role": "user", "content": f"Extract entities from this text:\n\n{text}"}
-    ]
-    
-    payload = json.dumps({
-        "model": model,
-        "messages": messages,
-        "temperature": 0.0,
-        "max_tokens": 2048
-    }).encode('utf-8')
-    
-    req = urllib.request.Request(
-        f"{api_base}/chat/completions",
-        data=payload,
-        headers={"Authorization": f"Bearer {api_key}", "Content-Type": "application/json"},
-        method="POST"
-    )
-    
-    try:
-        with urllib.request.urlopen(req, timeout=60) as resp:
-            result = json.loads(resp.read().decode('utf-8'))
-            content = result["choices"][0]["message"]["content"]
-            return parse_response(content)
-    except Exception as e:
-        print(f"ERROR: LLM call failed: {e}", file=sys.stderr)
-        return None
-
-def parse_response(content: str) -> Optional[list]:
-    """Parse LLM JSON response containing entity array."""
-    try:
-        data = json.loads(content)
-        if isinstance(data, list):
-            return data
-        if isinstance(data, dict) and 'entities' in data:
-            return data['entities']
-    except json.JSONDecodeError:
-        pass
-    import re
-    match = re.search(r'```(?:json)?\s*(\[.*?\])\s*```', content, re.DOTALL)
-    if match:
-        try:
-            data = json.loads(match.group(1))
-            if isinstance(data, list):
-                return data
-        except json.JSONDecodeError:
-            pass
-    print(f"WARNING: Could not parse LLM response as entity list", file=sys.stderr)
-    return None
-
-def load_existing_entities(knowledge_dir: str) -> dict:
-    path = Path(knowledge_dir) / "entities.json"
-    if not path.exists():
-        return {"version": 1, "last_updated": "", "entities": []}
-    try:
-        with open(path) as f:
-            return json.load(f)
-    except (json.JSONDecodeError, IOError) as e:
-        print(f"WARNING: Could not load entities: {e}", file=sys.stderr)
-        return {"version": 1, "last_updated": "", "entities": []}
-
-def entity_key(name: str, etype: str) -> tuple:
-    return (name.lower().strip(), etype.lower().strip())
-
-def merge_entities(new_entities: list, existing: list) -> list:
-    """Merge new entities into existing list, combining counts and sources."""
-    existing_by_key = {}
-    for e in existing:
-        key = entity_key(e.get('name',''), e.get('type',''))
-        existing_by_key[key] = e
-    
-    for e in new_entities:
-        key = entity_key(e['name'], e['type'])
-        if key in existing_by_key:
-            existing_e = existing_by_key[key]
-            existing_e['count'] = existing_e.get('count', 1) + 1
-            # Merge sources
-            old_sources = set(existing_e.get('sources', []))
-            new_sources = set(e.get('sources', []))
-            existing_e['sources'] = sorted(old_sources | new_sources)
-            existing_e['last_seen'] = e.get('last_seen', existing_e.get('last_seen'))
-        else:
-            e['count'] = e.get('count', 1)
-            e.setdefault('sources', [])
-            e.setdefault('first_seen', datetime.now(timezone.utc).isoformat())
-            existing.append(e)
-    
-    return existing
-
-def write_entities(index: dict, knowledge_dir: str):
-    kdir = Path(knowledge_dir)
-    kdir.mkdir(parents=True, exist_ok=True)
-    index['last_updated'] = datetime.now(timezone.utc).isoformat()
-    path = kdir / "entities.json"
-    with open(path, 'w', encoding='utf-8') as f:
-        json.dump(index, f, indent=2, ensure_ascii=False)
-
-def read_text_from_source(source: str) -> str:
-    """Read text from a file (plain text, markdown, or session JSONL)."""
-    path = Path(source)
-    if not path.exists():
-        raise FileNotFoundError(source)
-    if path.suffix == '.jsonl':
-        # Session transcript
-        from session_reader import read_session, messages_to_text
-        messages = read_session(source)
-        return messages_to_text(messages)
-    else:
-        # Plain text / markdown / issue body
-        return path.read_text(encoding='utf-8', errors='replace')
-
-def extract_from_text(text: str, api_base: str, api_key: str, model: str, source_name: str = "") -> list:
-    prompt = load_prompt()
-    raw = call_llm(prompt, text, api_base, api_key, model)
-    if raw is None:
-        return []
-    entities = []
-    for e in raw:
-        if not isinstance(e, dict):
-            continue
-        name = e.get('name', '').strip()
-        etype = e.get('type', '').strip().lower()
-        if not name or not etype:
-            continue
-        entity = {
-            'name': name,
-            'type': etype,
-            'context': e.get('context', '')[:200],
-            'last_seen': datetime.now(timezone.utc).isoformat(),
-            'sources': [source_name] if source_name else []
-        }
-        entities.append(entity)
-    return entities
-
-def main():
-    parser = argparse.ArgumentParser(description="Extract named entities from text sources")
-    parser.add_argument('--file', help='Single file to process')
-    parser.add_argument('--dir', help='Directory of files to process')
-    parser.add_argument('--session', help='Single session JSONL file')
-    parser.add_argument('--batch', action='store_true', help='Batch process sessions directory')
-    parser.add_argument('--sessions-dir', default=os.path.expanduser('~/.hermes/sessions'),
-                        help='Sessions directory for batch mode')
-    parser.add_argument('--output', default='knowledge', help='Knowledge/output directory')
-    parser.add_argument('--api-base', default=DEFAULT_API_BASE)
-    parser.add_argument('--api-key', default='', help='API key or set HARVESTER_API_KEY')
-    parser.add_argument('--model', default=DEFAULT_MODEL)
-    parser.add_argument('--dry-run', action='store_true', help='Preview without writing')
-    parser.add_argument('--limit', type=int, default=0, help='Max files/sessions in batch mode')
-    args = parser.parse_args()
-
-    api_key = args.api_key or DEFAULT_API_KEY or find_api_key()
-    if not api_key:
-        print("ERROR: No API key found", file=sys.stderr)
-        sys.exit(1)
-
-    knowledge_dir = args.output
-    if not os.path.isabs(knowledge_dir):
-        knowledge_dir = str(SCRIPT_DIR.parent / knowledge_dir)
-
-    sources = []
-    if args.file:
-        sources = [args.file]
-    elif args.dir:
-        files = sorted(Path(args.dir).rglob("*"))
-        sources = [str(f) for f in files if f.is_file() and f.suffix in ('.txt','.md','.json','.jsonl','.yaml','.yml')]
-        if args.limit > 0:
-            sources = sources[:args.limit]
-    elif args.session:
-        sources = [args.session]
-    elif args.batch:
-        sess_dir = Path(args.sessions_dir)
-        sources = sorted(sess_dir.glob("*.jsonl"), reverse=True)
-        if args.limit > 0:
-            sources = sources[:args.limit]
-        sources = [str(s) for s in sources]
-    else:
-        parser.print_help()
-        sys.exit(1)
-
-    print(f"Processing {len(sources)} sources...")
-    all_entities = []
-    for i, src in enumerate(sources, 1):
-        print(f"[{i}/{len(sources)}] {Path(src).name}...", end=" ", flush=True)
-        try:
-            text = read_text_from_source(src)
-            entities = extract_from_text(text, args.api_base, api_key, args.model, source_name=Path(src).name)
-            all_entities.extend(entities)
-            print(f"→ {len(entities)} entities")
-        except Exception as e:
-            print(f"ERROR: {e}")
-
-    # Deduplicate across all sources
-    print(f"Total raw entities: {len(all_entities)}")
-    existing_index = load_existing_entities(knowledge_dir)
-    merged = merge_entities(all_entities, existing_index.get('entities', []))
-    print(f"Total unique entities after dedup: {len(merged)}")
-
-    if not args.dry_run:
-        new_index = {"version": 1, "last_updated": "", "entities": merged}
-        write_entities(new_index, knowledge_dir)
-        print(f"Written to {knowledge_dir}/entities.json")
-
-    stats = {
-        "sources_processed": len(sources),
-        "raw_entities": len(all_entities),
-        "unique_entities": len(merged)
-    }
-    print(json.dumps(stats, indent=2))
-
-if __name__ == '__main__':
-    main()

scripts/genome_diff.py

@@ -0,0 +1,288 @@
+#!/usr/bin/env python3
+"""
+Codebase Genome Diff — Detect structural changes between two versions.
+
+Compares two git refs (commits, branches, tags) and produces a human-readable
+report of structural changes:
+  • Added/removed/renamed files
+  • Changed functions/classes (signature modifications)
+  • New dependencies (imports, requirements, etc.)
+
+Usage:
+    python3 scripts/genome_diff.py --ref1 <commit1> --ref2 <commit2>
+    python3 scripts/genome_diff.py --ref1 main --ref2 feature-branch
+    python3 scripts/genome_diff.py --ref1 v1.0 --ref2 v2.0 --output report.txt
+"""
+
+import argparse
+import json
+import os
+import re
+import subprocess
+import sys
+from dataclasses import dataclass, field
+from typing import List, Dict, Any, Optional
+
+SCRIPT_DIR = os.path.dirname(os.path.abspath(__file__))
+sys.path.insert(0, SCRIPT_DIR)
+from diff_analyzer import DiffAnalyzer, ChangeCategory
+
+
+@dataclass
+class FunctionChange:
+    file: str
+    name: str
+    kind: str  # 'function' or 'class'
+    change_type: str  # 'added' or 'removed' (simplified)
+    old_line: Optional[int] = None
+    new_line: Optional[int] = None
+
+
+@dataclass
+class DependencyChange:
+    file: str
+    module: str
+    change_type: str  # 'added' or 'removed' or 'modified'
+    line: int = 0
+
+
+@dataclass
+class GenomeDiffReport:
+    ref1: str
+    ref2: str
+    file_changes: List[Dict[str, Any]] = field(default_factory=list)
+    function_changes: List[FunctionChange] = field(default_factory=list)
+    dependency_changes: List[DependencyChange] = field(default_factory=list)
+    total_files_changed: int = 0
+    total_functions_changed: int = 0
+    total_dependencies_changed: int = 0
+
+    def to_dict(self) -> Dict[str, Any]:
+        return {
+            "ref1": self.ref1,
+            "ref2": self.ref2,
+            "summary": {
+                "files": self.total_files_changed,
+                "functions": self.total_functions_changed,
+                "dependencies": self.total_dependencies_changed,
+            },
+            "file_changes": self.file_changes,
+            "function_changes": [fc.__dict__ for fc in self.function_changes],
+            "dependency_changes": [dc.__dict__ for dc in self.dependency_changes],
+        }
+
+    def human_report(self) -> str:
+        lines = []
+        lines.append(f"Codebase Genome Diff: {self.ref1} → {self.ref2}")
+        lines.append("=" * 60)
+        lines.append(f"  Files changed: {self.total_files_changed}")
+        lines.append(f"  Functions changed: {self.total_functions_changed}")
+        lines.append(f"  Dependencies changed: {self.total_dependencies_changed}")
+        lines.append("")
+
+        for fc in self.file_changes:
+            kind = []
+            if fc.get('is_new'):
+                kind.append("NEW")
+            if fc.get('is_deleted'):
+                kind.append("DELETED")
+            if fc.get('is_renamed'):
+                kind.append("RENAMED")
+            if fc.get('is_binary'):
+                kind.append("BINARY")
+            kind_str = f" [{', '.join(kind)}]" if kind else ""
+            lines.append(f"  {fc['path']}{kind_str}  (+{fc['added_lines']}/-{fc['deleted_lines']})")
+        lines.append("")
+
+        for fc in self.function_changes:
+            op = {'added': '+', 'removed': '-', 'modified': '~'}.get(fc.change_type, '?')
+            lines.append(f"  [{op}] {fc.file}: {fc.kind} '{fc.name}'")
+        lines.append("")
+
+        for dc in self.dependency_changes:
+            op = '+' if dc.change_type == 'added' else '-'
+            lines.append(f"  [{op}] {dc.file}: {dc.module}")
+        lines.append("")
+
+        return "\n".join(lines)
+
+
+def run_git_diff(ref1: str, ref2: str) -> str:
+    result = subprocess.run(
+        ['git', 'diff', '--unified=0', f'{ref1}...{ref2}'],
+        capture_output=True, text=True, cwd=SCRIPT_DIR
+    )
+    if result.returncode not in (0, 1):
+        print(f"git diff failed: {result.stderr}", file=sys.stderr)
+        sys.exit(1)
+    return result.stdout
+
+
+def extract_function_changes(diff_text: str) -> List[FunctionChange]:
+    changes: List[FunctionChange] = []
+    pattern = re.compile(r'^([+\-])\s*(def|class)\s+(\w+)', re.MULTILINE)
+    hunk_header_re = re.compile(r'^@@\s+-(\d+)(?:,(\d+))?\s+\+(\d+)(?:,(\d+))?\s+@@')
+    current_old_line: Optional[int] = None
+    current_new_line: Optional[int] = None
+
+    for line in diff_text.split('\n'):
+        hdr = hunk_header_re.match(line)
+        if hdr:
+            current_old_line = int(hdr.group(1))
+            current_new_line = int(hdr.group(3))
+            continue
+        m = pattern.match(line)
+        if m:
+            op = m.group(1)
+            kind = m.group(2)
+            name = m.group(3)
+            change_type = "added" if op == '+' else "removed"
+            line_num = current_new_line if change_type == "added" else current_old_line
+            changes.append(FunctionChange(
+                file="<unknown>",
+                name=name,
+                kind=kind,
+                change_type=change_type,
+                new_line=line_num if change_type == "added" else None,
+                old_line=line_num if change_type == "removed" else None,
+            ))
+            # Advance line counters heuristically
+            if op == '-':
+                if current_old_line is not None:
+                    current_old_line += 1
+            elif op == '+':
+                if current_new_line is not None:
+                    current_new_line += 1
+        elif line.startswith(' '):
+            if current_old_line is not None:
+                current_old_line += 1
+            if current_new_line is not None:
+                current_new_line += 1
+        # lines starting with other prefixes (like \\ No newline) ignored
+    return changes
+
+
+def extract_dependency_changes(diff_text: str, analyzer: DiffAnalyzer) -> List[DependencyChange]:
+    changes: List[DependencyChange] = []
+    import_pattern = re.compile(
+        r'^([+\-])\s*(?:import\s+([\w\.]+)|from\s+([\w\.]+)\s+import)',
+        re.MULTILINE
+    )
+    file_diffs = analyzer._split_files(diff_text)
+    for file_diff in file_diffs:
+        file_match = re.search(r'^diff --git a/.*? b/(.*?)$', file_diff, re.MULTILINE)
+        if not file_match:
+            continue
+        filepath = file_match.group(1)
+
+        # Scan each line for import changes
+        for line in file_diff.split('\n'):
+            m = import_pattern.match(line)
+            if m:
+                change_type = "added" if m.group(1) == '+' else "removed"
+                module = m.group(2) or m.group(3)
+                changes.append(DependencyChange(
+                    file=filepath,
+                    module=module,
+                    change_type=change_type,
+                    line=0
+                ))
+
+        # Detect if this file is a dependency manifest
+        req_file_pattern = re.compile(
+            r'^[\+\-].*?(requirements(.*?)\.txt|pyproject\.toml|setup\.py|Pipfile)'
+        )
+        if any(req_file_pattern.match(line) for line in file_diff.split('\n')):
+            if not any(c.file == filepath and c.module == "<file>" for c in changes):
+                changes.append(DependencyChange(
+                    file=filepath,
+                    module="<file>",
+                    change_type="modified",
+                    line=0
+                ))
+    return changes
+
+
+def correlate_function_changes_with_files(diff_text: str, functions: List[FunctionChange]) -> List[FunctionChange]:
+    result: List[FunctionChange] = []
+    # Split diff into per-file sections
+    file_sections: List[tuple[str, str]] = []
+    current_file: Optional[str] = None
+    current_lines: List[str] = []
+    for line in diff_text.split('\n'):
+        if line.startswith('diff --git'):
+            if current_file is not None:
+                file_sections.append((current_file, '\n'.join(current_lines)))
+            m = re.match(r'^diff --git a/.*? b/(.*?)$', line)
+            current_file = m.group(1) if m else "unknown"
+            current_lines = [line]
+        else:
+            current_lines.append(line)
+    if current_file is not None:
+        file_sections.append((current_file, '\n'.join(current_lines)))
+
+    pattern = re.compile(r'^([+\-])\s*(def|class)\s+(\w+)', re.MULTILINE)
+    for filepath, section in file_sections:
+        for m in pattern.finditer(section):
+            op = m.group(1)
+            kind = m.group(2)
+            name = m.group(3)
+            change_type = "added" if op == '+' else "removed"
+            result.append(FunctionChange(
+                file=filepath,
+                name=name,
+                kind=kind,
+                change_type=change_type
+            ))
+    return result
+
+
+def main():
+    parser = argparse.ArgumentParser(description="Codebase Genome Diff — structural changes between versions")
+    parser.add_argument("--ref1", required=True, help="First git ref (commit, branch, tag)")
+    parser.add_argument("--ref2", required=True, help="Second git ref")
+    parser.add_argument("--output", help="Write report to file")
+    parser.add_argument("--json", action="store_true", help="Output JSON instead of human report")
+    args = parser.parse_args()
+
+    try:
+        diff_text = run_git_diff(args.ref1, args.ref2)
+    except Exception as e:
+        print(f"Error: {e}", file=sys.stderr)
+        sys.exit(1)
+
+    if not diff_text.strip():
+        print(f"No differences between {args.ref1} and {args.ref2}.")
+        sys.exit(0)
+
+    analyzer = DiffAnalyzer()
+    summary = analyzer.analyze(diff_text)
+
+    file_changes = [fc.to_dict() for fc in summary.files]
+    func_changes = extract_function_changes(diff_text)
+    func_changes = correlate_function_changes_with_files(diff_text, func_changes)
+    dep_changes = extract_dependency_changes(diff_text, analyzer)
+
+    report = GenomeDiffReport(
+        ref1=args.ref1,
+        ref2=args.ref2,
+        file_changes=file_changes,
+        function_changes=func_changes,
+        dependency_changes=dep_changes,
+        total_files_changed=len(file_changes),
+        total_functions_changed=len(func_changes),
+        total_dependencies_changed=len(dep_changes),
+    )
+
+    output = json.dumps(report.to_dict(), indent=2) if args.json else report.human_report()
+
+    if args.output:
+        with open(args.output, 'w') as f:
+            f.write(output + '\n')
+        print(f"Report written to {args.output}")
+    else:
+        print(output)
+
+
+if __name__ == '__main__':
+    main()

scripts/test_entity_extractor.py

@@ -1,116 +0,0 @@
-#!/usr/bin/env python3
-"""
-Smoke test for entity_extractor pipeline — verifies:
-- session/plain text reading
-- mock LLM entity extraction
-- deduplication and merging
-- output file format
-
-Does NOT call the real LLM.
-"""
-
-import json
-import os
-import tempfile
-from unittest.mock import patch
-import sys
-from pathlib import Path
-
-SCRIPT_DIR = Path(__file__).parent.absolute()
-sys.path.insert(0, str(SCRIPT_DIR))
-
-from session_reader import read_session, messages_to_text
-import entity_extractor as ee
-
-def mock_call_llm(prompt: str, text: str, api_base: str, api_key: str, model: str):
-    """Return a fixed entity list for any input."""
-    return [
-        {"name": "Hermes", "type": "tool", "context": "Hermes agent uses the tools tool."},
-        {"name": "Gitea", "type": "tool", "context": "Gitea is a forge."},
-        {"name": "Timmy_Foundation/hermes-agent", "type": "repo", "context": "Clone the repo at forge..."},
-    ]
-
-def test_read_session_text():
-    with tempfile.NamedTemporaryFile(mode='w', suffix='.jsonl', delete=False) as f:
-        f.write('{"role": "user", "content": "Clone repo", "timestamp": "2026-04-13T10:00:00Z"}\n')
-        f.write('{"role": "assistant", "content": "Done", "timestamp": "2026-04-13T10:00:05Z"}\n')
-        path = f.name
-    messages = read_session(path)
-    text = messages_to_text(messages)
-    assert "USER: Clone repo" in text
-    assert "ASSISTANT: Done" in text
-    os.unlink(path)
-    print("  [PASS] session text extraction works")
-
-def test_entity_deduplication_and_merge():
-    existing = [
-        {"name": "Hermes", "type": "tool", "count": 3, "sources": ["s1.jsonl"]}
-    ]
-    new = [
-        {"name": "Hermes", "type": "tool", "sources": ["s2.jsonl"]},
-        {"name": "Gitea", "type": "tool", "sources": ["s2.jsonl"]},
-    ]
-    merged = ee.merge_entities(new, existing.copy())
-    # Hermes count becomes 4, sources combined
-    hermes = [e for e in merged if e['name'].lower() == 'hermes'][0]
-    assert hermes['count'] == 4
-    assert set(hermes['sources']) == {'s1.jsonl', 's2.jsonl'}
-    # Gitea new entry
-    gitea = [e for e in merged if e['name'].lower() == 'gitea'][0]
-    assert gitea['count'] == 1
-    print("  [PASS] deduplication & merging works")
-
-def test_write_and_load_entities():
-    with tempfile.TemporaryDirectory() as tmp:
-        kdir = Path(tmp) / "knowledge"
-        kdir.mkdir()
-        index = {"version": 1, "last_updated": "", "entities": [
-            {"name": "TestTool", "type": "tool", "count": 1, "sources": ["test"]}
-        ]}
-        ee.write_entities(index, str(kdir))
-        # load back
-        loaded = ee.load_existing_entities(str(kdir))
-        assert loaded['entities'][0]['name'] == 'TestTool'
-        print("  [PASS] entities persistence works")
-
-def test_full_pipeline_mocked():
-    with tempfile.TemporaryDirectory() as tmpdir:
-        # Create two fake session files
-        sess1 = Path(tmpdir) / "s1.jsonl"
-        sess1.write_text('{"role":"user","content":"Use Hermes to clone","timestamp":"..."}\n')
-        sess2 = Path(tmpdir) / "s2.jsonl"
-        sess2.write_text('{"role":"user","content":"Deploy with Gitea","timestamp":"..."}\n')
-        
-        knowledge_dir = Path(tmpdir) / "knowledge"
-        knowledge_dir.mkdir()
-        
-        # Patch call_llm
-        with patch('entity_extractor.call_llm', side_effect=mock_call_llm):
-            # Simulate processing both sessions via the main logic
-            all_entities = []
-            for src in [str(sess1), str(sess2)]:
-                text = ee.read_text_from_source(src)
-                ents = ee.extract_from_text(text, "http://api", "fake-key", "model", source_name=Path(src).name)
-                all_entities.extend(ents)
-            
-            # Merge into empty index
-            merged = ee.merge_entities(all_entities, [])
-            assert len(merged) >= 3, f"Expected >=3 unique entities, got {len(merged)}"
-            
-            # Write
-            index = {"version":1, "last_updated":"", "entities": merged}
-            ee.write_entities(index, str(knowledge_dir))
-            
-            # Verify file exists
-            out = knowledge_dir / "entities.json"
-            assert out.exists()
-            data = json.loads(out.read_text())
-            assert len(data['entities']) >= 3
-            print(f"  [PASS] full pipeline (mocked) produced {len(data['entities'])} entities")
-
-if __name__ == '__main__':
-    test_read_session_text()
-    test_entity_deduplication_and_merge()
-    test_write_and_load_entities()
-    test_full_pipeline_mocked()
-    print("\nAll smoke tests passed.")

templates/entity-extraction-prompt.md

@@ -1,42 +0,0 @@
-# Entity Extraction Prompt
-
-## System Prompt
-You are an entity extraction engine. You read text and output ONLY a JSON array of named entities. You do not infer. You extract only what the text explicitly mentions.
-
-## Task
-Extract all named entities from the provided text. Categorize each entity into exactly one of these types:
-- `person` — individual's name (e.g., Alexander, Rockachopa, Allegro)
-- `project` — software project or component name (e.g., The Nexus, Timmy Home, compounding-intelligence)
-- `tool` — software tool, command, library, framework (e.g., git, Docker, PyTorch, Hermes)
-- `concept` — abstract idea, methodology, paradigm (e.g., compounding intelligence, bootstrap, harvester)
-- `repo` — repository reference in the form `owner/repo` or URL pointing to a repo
-
-## Rules
-1. Extract ONLY names that appear explicitly in the text.
-2. Do NOT infer, assume, or hallucinate.
-3. Each entity must have: `name` (exact string), `type` (one of the five above), and `context` (short snippet showing usage, 1-2 sentences).
-4. The same entity mentioned multiple times should appear only ONCE in the output (deduplicate by name+type).
-5. For `repo` type, match patterns like `owner/repo`, `github.com/owner/repo`, `forge.alexanderwhitestone.com/owner/repo`.
-6. For `tool` type, include commands (git, pytest), platforms (Linux, macOS), runtimes (Python, Node.js), and CLI utilities.
-7. For `person` type, look for capitalized full names, or single names used in personal attribution ("asked Alex", "for Alexander").
-8. For `concept`, include technical terms that represent an idea rather than a concrete thing.
-
-## Output Format
-Return ONLY valid JSON, no markdown, no explanation. Array of objects:
-```json
-[
-  {
-    "name": "Hermes",
-    "type": "tool",
-    "context": "Hermes agent uses the tools tool to execute commands."
-  },
-  {
-    "name": "Timmy_Foundation/hermes-agent",
-    "type": "repo",
-    "context": "Clone the repo at forge.../Timmy_Foundation/hermes-agent"
-  }
-]
-```
-
-## Text to extract from:
-{{text}}

tests/test_entity_extractor.py

@@ -1,82 +0,0 @@
-"""
-Test suite for entity_extractor.py (Issue #144).
-
-Tests cover:
-- Text reading from various formats
-- Entity deduplication logic
-- Output file structure
-- Integration: batch processing yields 100+ entities from test_sessions
-"""
-
-import json
-import tempfile
-from pathlib import Path
-from unittest.mock import patch, MagicMock
-
-# We'll test the pure functions directly; avoid hitting real LLM in unit tests
-import sys
-sys.path.insert(0, str(Path(__file__).resolve().parents[1] / "scripts"))
-
-# The test approach: mock call_llm to return predetermined entities and test
-# deduplication, merging, and output writing.
-
-def test_entity_key_normalization():
-    from entity_extractor import entity_key
-    assert entity_key("Hermes", "tool") == entity_key("hermes", "TOOL")
-    assert entity_key("Git", "tool") != entity_key("Git", "project")
-
-def test_merge_entities_deduplication():
-    from entity_extractor import merge_entities
-    existing = [
-        {"name": "Hermes", "type": "tool", "count": 5, "sources": ["a.jsonl"]}
-    ]
-    new = [
-        {"name": "Hermes", "type": "tool", "sources": ["b.jsonl"]},
-        {"name": "Gitea", "type": "tool", "sources": ["b.jsonl"]}
-    ]
-    merged = merge_entities(new, existing.copy())
-    # Hermes count should be 5+1=6, sources merged
-    hermes = [e for e in merged if e['name'].lower()=='hermes'][0]
-    assert hermes['count'] == 6
-    assert set(hermes['sources']) == {"a.jsonl", "b.jsonl"}
-    # Gitea added fresh
-    gitea = [e for e in merged if e['name'].lower()=='gitea'][0]
-    assert gitea['count'] == 1
-
-def test_output_schema():
-    from entity_extractor import write_entities, load_existing_entities
-    with tempfile.TemporaryDirectory() as tmp:
-        kdir = Path(tmp) / "knowledge"
-        kdir.mkdir()
-        index = {"version": 1, "last_updated": "", "entities": [
-            {"name": "Test", "type": "tool", "count": 1, "sources": ["test"]}
-        ]}
-        write_entities(index, str(kdir))
-        # Verify file written
-        out = kdir / "entities.json"
-        assert out.exists()
-        data = json.loads(out.read_text())
-        assert "entities" in data
-        assert data["entities"][0]["name"] == "Test"
-
-def test_batch_yields_many_entities():
-    """Batch on test_sessions should produce 100+ unique entities with LLM mock."""
-    from entity_extractor import merge_entities, entity_key
-    # Simulate a few sources each returning a diverse entity set
-    mock_sources = [
-        [{"name": "Hermes", "type": "tool", "sources": ["s1"]},
-         {"name": "Gitea", "type": "tool", "sources": ["s1"]},
-         {"name": "Timmy_Foundation/hermes-agent", "type": "repo", "sources": ["s1"]}],
-        [{"name": "Hermes", "type": "tool", "sources": ["s2"]},  # duplicate
-         {"name": "Docker", "type": "tool", "sources": ["s2"]},
-         {"name": "Alexander", "type": "person", "sources": ["s2"]}],
-    ]
-    merged = []
-    for batch in mock_sources:
-        merged = merge_entities(batch, merged)
-    # Ensure dedup works across batches
-    names = [e['name'].lower() for e in merged]
-    assert names.count('hermes') == 1
-    assert len(merged) == 4  # Hermes, Gitea, repo, Docker, Alexander
-
-# The real LLM extraction test would require live API key; skip in CI