feat: add dependency freshness checker — issue #161

Implements scripts/dependency_freshness.py which compares installed
dependencies against latest PyPI versions and flags packages that are
more than 2 major versions behind. Includes comprehensive tests in
scripts/test_dependency_freshness.py.

Closes #161

scripts/dependency_freshness.py

@@ -0,0 +1,271 @@
+#!/usr/bin/env python3
+"""dependency_freshness.py - Compare installed dependencies against latest PyPI versions.
+
+Identify packages that are more than 2 major versions behind.
+Outputs a human-readable report by default or JSON with --json flag.
+"""
+
+import argparse
+import json
+import subprocess
+import sys
+from packaging import version
+from typing import Dict, List, Tuple
+
+
+def parse_requirements(requirements_path: str) -> List[str]:
+    """Parse package names from a requirements.txt file."""
+    packages = []
+    try:
+        with open(requirements_path, 'r') as f:
+            for line in f:
+                line = line.strip()
+                if not line or line.startswith('#'):
+                    continue
+                pkg_name = line
+                for delim in ['[', '>', '<', '=', '!', ';', '@']:
+                    if delim in pkg_name:
+                        pkg_name = pkg_name.split(delim)[0]
+                pkg_name = pkg_name.strip()
+                if pkg_name:
+                    packages.append(pkg_name.lower())
+    except FileNotFoundError:
+        print(f"Warning: requirements file not found: {requirements_path}", file=sys.stderr)
+    return packages
+
+
+def get_installed_packages() -> Dict[str, str]:
+    """Get all installed packages via pip list --format=json."""
+    try:
+        result = subprocess.run(
+            [sys.executable, '-m', 'pip', 'list', '--format=json'],
+            capture_output=True, text=True, check=True
+        )
+        packages = json.loads(result.stdout)
+        return {pkg['name'].lower(): pkg['version'] for pkg in packages}
+    except subprocess.CalledProcessError as e:
+        print(f"Error running pip list: {e}", file=sys.stderr)
+        sys.exit(1)
+    except json.JSONDecodeError as e:
+        print(f"Error parsing pip output: {e}", file=sys.stderr)
+        sys.exit(1)
+
+
+def get_outdated_packages() -> Dict[str, dict]:
+    """Get outdated packages via pip list --outdated --format=json."""
+    try:
+        result = subprocess.run(
+            [sys.executable, '-m', 'pip', 'list', '--outdated', '--format=json'],
+            capture_output=True, text=True, check=True
+        )
+        outdated_list = json.loads(result.stdout)
+        outdated = {}
+        for pkg in outdated_list:
+            name = pkg['name'].lower()
+            outdated[name] = {
+                'installed': pkg.get('version', ''),
+                'latest': pkg.get('latest_version', ''),
+                'latest_filetype': pkg.get('latest_filetype', '')
+            }
+        return outdated
+    except subprocess.CalledProcessError as e:
+        print(f"Error running pip list --outdated: {e}", file=sys.stderr)
+        sys.exit(1)
+    except json.JSONDecodeError as e:
+        print(f"Error parsing pip outdated output: {e}", file=sys.stderr)
+        sys.exit(1)
+
+
+def get_major_version(v: str) -> int:
+    """Extract major version number from a version string."""
+    try:
+        parsed = version.parse(v)
+        if hasattr(parsed, 'major'):
+            return int(parsed.major)
+        parts = str(v).split('.')
+        if parts:
+            return int(parts[0])
+    except Exception:
+        pass
+    return 0
+
+
+def is_more_than_two_majors_behind(installed_ver: str, latest_ver: str) -> bool:
+    """Check if installed version is more than 2 major versions behind latest."""
+    try:
+        installed_major = get_major_version(installed_ver)
+        latest_major = get_major_version(latest_ver)
+        return (latest_major - installed_major) > 2
+    except Exception:
+        return False
+
+
+def analyze_dependencies(
+    required_packages: List[str],
+    installed_packages: Dict[str, str],
+    outdated_packages: Dict[str, dict]
+) -> Tuple[List[dict], List[str], List[dict]]:
+    """Analyze dependency freshness."""
+    very_outdated = []
+    missing = []
+    outdated_but_not_critical = []
+
+    for pkg in required_packages:
+        if pkg not in installed_packages:
+            missing.append(pkg)
+            continue
+
+        installed_ver = installed_packages[pkg]
+        if pkg not in outdated_packages:
+            continue
+
+        latest_ver = outdated_packages[pkg]['latest']
+        if is_more_than_two_majors_behind(installed_ver, latest_ver):
+            very_outdated.append({
+                'package': pkg,
+                'installed': installed_ver,
+                'latest': latest_ver,
+                'major_diff': get_major_version(latest_ver) - get_major_version(installed_ver)
+            })
+        else:
+            outdated_but_not_critical.append({
+                'package': pkg,
+                'installed': installed_ver,
+                'latest': latest_ver,
+                'major_diff': get_major_version(latest_ver) - get_major_version(installed_ver)
+            })
+
+    return very_outdated, missing, outdated_but_not_critical
+
+
+def generate_human_report(
+    very_outdated: List[dict],
+    missing: List[str],
+    outdated_but_not_critical: List[dict],
+    requirements_path: str
+) -> str:
+    """Generate a human-readable staleness report."""
+    lines = []
+    lines.append("=" * 60)
+    lines.append("DEPENDENCY FRESHNESS REPORT")
+    lines.append("=" * 60)
+    lines.append(f"Requirements file: {requirements_path}")
+    total = len(very_outdated) + len(missing) + len(outdated_but_not_critical)
+    lines.append(f"Total dependencies checked: {total}")
+    lines.append(f"Very outdated (>2 major versions behind): {len(very_outdated)}")
+    lines.append(f"Outdated but within 2 major versions: {len(outdated_but_not_critical)}")
+    lines.append(f"Missing (not installed): {len(missing)}")
+    lines.append("")
+
+    if very_outdated:
+        lines.append("!!! VERY OUTDATED PACKAGES (consider updating):")
+        lines.append("-" * 60)
+        for pkg_info in very_outdated:
+            lines.append(f"  {pkg_info['package']}")
+            lines.append(f"    Installed: {pkg_info['installed']}")
+            lines.append(f"    Latest:    {pkg_info['latest']}")
+            lines.append(f"    Major diff: {pkg_info['major_diff']}")
+            lines.append("")
+    else:
+        lines.append("✓ No packages more than 2 major versions behind.")
+        lines.append("")
+
+    if outdated_but_not_critical:
+        lines.append(f"Outdated packages (within 2 major versions):")
+        lines.append("-" * 60)
+        for pkg_info in outdated_but_not_critical:
+            lines.append(f"  {pkg_info['package']}: {pkg_info['installed']} -> {pkg_info['latest']} (major diff: {pkg_info['major_diff']})")
+        lines.append("")
+
+    if missing:
+        lines.append(f"Missing packages (not installed):")
+        lines.append("-" * 60)
+        for pkg in missing:
+            lines.append(f"  {pkg}")
+        lines.append("")
+
+    lines.append("=" * 60)
+    lines.append("For full details, run: python3 -m pip list --outdated")
+    lines.append("=" * 60)
+
+    return "\n".join(lines)
+
+
+def generate_json_report(
+    very_outdated: List[dict],
+    missing: List[str],
+    outdated_but_not_critical: List[dict],
+    requirements_path: str
+) -> str:
+    """Generate a JSON staleness report."""
+    report = {
+        'requirements_file': requirements_path,
+        'summary': {
+            'total_dependencies': len(very_outdated) + len(missing) + len(outdated_but_not_critical),
+            'very_outdated_count': len(very_outdated),
+            'outdated_within_threshold_count': len(outdated_but_not_critical),
+            'missing_count': len(missing)
+        },
+        'very_outdated': very_outdated,
+        'outdated_within_threshold': outdated_but_not_critical,
+        'missing': missing
+    }
+    return json.dumps(report, indent=2)
+
+
+def main():
+    parser = argparse.ArgumentParser(
+        description='Check dependency freshness against PyPI latest versions.'
+    )
+    parser.add_argument(
+        '--requirements', '-r',
+        default='requirements.txt',
+        help='Path to requirements.txt file (default: requirements.txt)'
+    )
+    parser.add_argument(
+        '--json',
+        action='store_true',
+        help='Output report as JSON instead of human-readable text'
+    )
+    parser.add_argument(
+        '--output', '-o',
+        help='Optional output file for the report (default: stdout)'
+    )
+
+    args = parser.parse_args()
+
+    # Parse requirements
+    required_packages = parse_requirements(args.requirements)
+    if not required_packages:
+        print("No packages found in requirements file.", file=sys.stderr)
+        sys.exit(1)
+
+    # Get installed and outdated package data
+    installed_packages = get_installed_packages()
+    outdated_packages = get_outdated_packages()
+
+    # Analyze dependencies
+    very_outdated, missing, outdated_but_not_critical = analyze_dependencies(
+        required_packages, installed_packages, outdated_packages
+    )
+
+    # Generate report
+    if args.json:
+        report = generate_json_report(very_outdated, missing, outdated_but_not_critical, args.requirements)
+    else:
+        report = generate_human_report(very_outdated, missing, outdated_but_not_critical, args.requirements)
+
+    # Output report
+    if args.output:
+        with open(args.output, 'w') as f:
+            f.write(report + '\n')
+    else:
+        print(report)
+
+    # Exit code: 0 if no very outdated deps, 1 otherwise
+    exit_code = 1 if very_outdated else 0
+    sys.exit(exit_code)
+
+
+if __name__ == '__main__':
+    main()

scripts/knowledge_synthesizer.py

@@ -1,418 +0,0 @@
-#!/usr/bin/env python3
-"""
-knowledge_synthesizer.py — Zero-shot knowledge synthesis for compounding intelligence.
-
-Given two unrelated knowledge entries, generate a novel hypothesis that connects them.
-Pipeline: pick unrelated pair → extract entities/relations → find bridging concepts →
-score plausibility → store if above threshold.
-
-Usage:
-    python3 scripts/knowledge_synthesizer.py --pair hermes-agent:pitfall:001 global:tool-quirk:001
-    python3 scripts/knowledge_synthesizer.py --auto --threshold 0.75
-    python3 scripts/knowledge_synthesizer.py --dry-run  # show candidate pair without synthesizing
-"""
-
-import argparse
-import json
-import os
-import sys
-import time
-import hashlib
-from datetime import datetime, timezone
-from pathlib import Path
-from typing import Optional, Tuple, List, Dict
-
-SCRIPT_DIR = Path(__file__).parent.absolute()
-sys.path.insert(0, str(SCRIPT_DIR))
-
-REPO_ROOT = SCRIPT_DIR.parent
-KNOWLEDGE_DIR = REPO_ROOT / "knowledge"
-TEMPLATE_PATH = SCRIPT_DIR.parent / "templates" / "synthesis-prompt.md"
-
-# Default API configuration
-DEFAULT_API_BASE = os.environ.get(
-    "SYNTHESIS_API_BASE",
-    os.environ.get("HARVESTER_API_BASE", "https://api.nousresearch.com/v1")
-)
-DEFAULT_API_KEY = os.environ.get("SYNTHESIS_API_KEY", "")
-DEFAULT_MODEL = os.environ.get(
-    "SYNTHESIS_MODEL",
-    os.environ.get("HARVESTER_MODEL", "xiaomi/mimo-v2-pro")
-)
-
-# Places to look for API keys if not in env
-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_index() -> dict:
-    index_path = KNOWLEDGE_DIR / "index.json"
-    if not index_path.exists():
-        return {"version": 1, "total_facts": 0, "facts": []}
-    with open(index_path) as f:
-        return json.load(f)
-
-
-def save_index(index: dict) -> None:
-    KNOWLEDGE_DIR.mkdir(parents=True, exist_ok=True)
-    index_path = KNOWLEDGE_DIR / "index.json"
-    with open(index_path, 'w', encoding='utf-8') as f:
-        json.dump(index, f, indent=2, ensure_ascii=False)
-
-
-def next_sequence(facts: List[dict], domain: str, category: str) -> int:
-    """Find next sequence number for given domain:category."""
-    prefix = f"{domain}:{category}:"
-    max_seq = 0
-    for fact in facts:
-        fid = fact.get('id', '')
-        if fid.startswith(prefix):
-            try:
-                seq = int(fid.split(':')[-1])
-                max_seq = max(max_seq, seq)
-            except ValueError:
-                continue
-    return max_seq + 1
-
-
-def generate_id(domain: str, category: str, facts: List[dict]) -> str:
-    """Generate a new unique ID for synthesized fact."""
-    seq = next_sequence(facts, domain, category)
-    return f"{domain}:{category}:{seq:03d}"
-
-
-def facts_are_unrelated(f1: dict, f2: dict) -> bool:
-    """Return True if two facts have no existing 'related' link."""
-    id1, id2 = f1['id'], f2['id']
-    rel1 = set(f1.get('related', []))
-    rel2 = set(f2.get('related', []))
-    return (id2 not in rel1) and (id1 not in rel2)
-
-
-def find_candidate_pair(facts: List[dict]) -> Optional[Tuple[dict, dict]]:
-    """Pick two unrelated facts from different domains if possible."""
-    # Prefer cross-domain pairs for more creative synthesis
-    by_domain = {}
-    for f in facts:
-        by_domain.setdefault(f['domain'], []).append(f)
-
-    domains = list(by_domain.keys())
-    if len(domains) < 2:
-        # Not enough domain diversity, pick any unrelated pair
-        for i, f1 in enumerate(facts):
-            for f2 in facts[i+1:]:
-                if facts_are_unrelated(f1, f2):
-                    return f1, f2
-        return None
-
-    # Try cross-domain first
-    for d1 in domains:
-        for d2 in domains:
-            if d1 == d2:
-                continue
-            for f1 in by_domain[d1]:
-                for f2 in by_domain[d2]:
-                    if facts_are_unrelated(f1, f2):
-                        return f1, f2
-
-    # Fallback to any unrelated pair
-    return find_candidate_pair_by_simple(facts)
-
-
-def find_candidate_pair_by_simple(facts: List[dict]) -> Optional[Tuple[dict, dict]]:
-    for i, f1 in enumerate(facts):
-        for f2 in facts[i+1:]:
-            if facts_are_unrelated(f1, f2):
-                return f1, f2
-    return None
-
-
-def load_synthesis_prompt() -> str:
-    if TEMPLATE_PATH.exists():
-        return TEMPLATE_PATH.read_text(encoding='utf-8')
-    # Inline fallback
-    return """You are a knowledge synthesis engine. Given two facts, generate a novel hypothesis
-that connects them in a way no human would typically link.
-
-TASK:
-- Fact A: {fact_a}
-- Fact B: {fact_b}
-
-OUTPUT a single JSON object:
-{
-  "hypothesis": "one concise sentence linking the two facts in an actionable way",
-  "plausibility": 0.0-1.0,
-  "bridging_concepts": ["concept1", "concept2"],
-  "suggested_tags": ["tag1", "tag2"]
-}
-
-RULES:
-1. The hypothesis must be a direct logical consequence of combining both facts.
-2. Do NOT restate either fact — produce a new insight.
-3. Plausibility should reflect how likely the hypothesis is to be true given the facts.
-4. If no meaningful connection exists, return {"hypothesis":"","plausibility":0.0}.
-5. Output ONLY valid JSON, no markdown.
-"""
-
-
-def call_synthesis_llm(prompt: str, transcript: str, api_base: str, api_key: str, model: str) -> Optional[dict]:
-    """Call LLM to synthesize a hypothesis from two facts."""
-    import urllib.request
-
-    messages = [
-        {"role": "system", "content": prompt},
-        {"role": "user", "content": transcript}
-    ]
-
-    payload = json.dumps({
-        "model": model,
-        "messages": messages,
-        "temperature": 0.7,  # More creative for synthesis
-        "max_tokens": 512
-    }).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_synthesis_response(content)
-    except Exception as e:
-        print(f"ERROR: LLM call failed: {e}", file=sys.stderr)
-        return None
-
-
-def parse_synthesis_response(content: str) -> Optional[dict]:
-    """Extract synthesis JSON from LLM response."""
-    try:
-        data = json.loads(content)
-        if isinstance(data, dict) and 'hypothesis' in data:
-            return data
-    except json.JSONDecodeError:
-        pass
-
-    import re
-    json_match = re.search(r'```(?:json)?\s*({.*?})\s*```', content, re.DOTALL)
-    if json_match:
-        try:
-            data = json.loads(json_match.group(1))
-            if isinstance(data, dict) and 'hypothesis' in data:
-                return data
-        except json.JSONDecodeError:
-            pass
-
-    # Try finding any JSON object
-    json_match = re.search(r'(\{.*"hypothesis".*\})', content, re.DOTALL)
-    if json_match:
-        try:
-            return json.loads(json_match.group(1))
-        except json.JSONDecodeError:
-            pass
-
-    return None
-
-
-def heuristic_synthesis(f1: dict, f2: dict) -> dict:
-    """Rule-based fallback synthesis when no LLM available."""
-    # Simple bridging: combine tags and domains
-    tags = list(set(f1.get('tags', []) + f2.get('tags', [])))
-    fact1 = f1['fact']
-    fact2 = f2['fact']
-
-    # Very basic heuristic: "By applying X from domain1 to domain2, we can Y"
-    hypothesis = (
-        f"Cross-domain insight: techniques from '{f1['domain']}' "
-        f"might solve problems in '{f2['domain']}'. "
-        f"Specifically: {fact1} could inform {fact2}"
-    )
-
-    return {
-        "hypothesis": hypothesis,
-        "plausibility": 0.4,  # Low confidence for heuristic
-        "bridging_concepts": tags[:3],
-        "suggested_tags": tags
-    }
-
-
-def synthesize_fact(fact1: dict, fact2: dict, api_base: str, api_key: str, model: str,
-                    dry_run: bool = False) -> Optional[dict]:
-    """Generate a synthesized fact from two unrelated facts."""
-    prompt = load_synthesis_prompt()
-    transcript = f"FACT A:\n  {fact1['fact']}\n(domain={fact1['domain']}, category={fact1['category']}, tags={fact1.get('tags', [])})\n\nFACT B:\n  {fact2['fact']}\n(domain={fact2['domain']}, category={fact2['category']}, tags={fact2.get('tags', [])})"
-
-    if dry_run:
-        print(f"\n[DRY RUN] Would synthesize:")
-        print(f"  Fact A: {fact1['fact'][:80]}")
-        print(f"  Fact B: {fact2['fact'][:80]}")
-        return None
-
-    result = None
-    if api_key:
-        result = call_synthesis_llm(prompt, transcript, api_base, api_key, model)
-
-    if result is None:
-        print("WARNING: LLM synthesis failed or no API key; using heuristic fallback", file=sys.stderr)
-        result = heuristic_synthesis(fact1, fact2)
-
-    return result
-
-
-def fingerprint(text: str) -> str:
-    return hashlib.md5(text.lower().strip().encode('utf-8')).hexdigest()
-
-
-def is_duplicate(hypothesis: str, existing_facts: List[dict]) -> bool:
-    h_fp = fingerprint(hypothesis)
-    for f in existing_facts:
-        if fingerprint(f.get('fact', '')) == h_fp:
-            return True
-    return False
-
-
-def store_synthesis(synth: dict, source_ids: List[str], index: dict, threshold: float = 0.5) -> bool:
-    """Store synthesized fact if plausibility exceeds threshold."""
-    plaus = synth.get('plausibility', 0.0)
-    if plaus < threshold:
-        print(f"Skipped: plausibility {plaus:.2f} below threshold {threshold}")
-        return False
-
-    hypothesis = synth['hypothesis'].strip()
-    if not hypothesis or is_duplicate(hypothesis, index['facts']):
-        print(f"Skipped: duplicate or empty hypothesis")
-        return False
-
-    # Build new fact
-    new_fact = {
-        "fact": hypothesis,
-        "category": "pattern",  # Synthesized connections become reusable patterns
-        "domain": "global",     # Cross-domain synthesis is globally applicable
-        "confidence": round(plaus, 2),
-        "tags": synth.get('suggested_tags', []),
-        "related": source_ids,
-        "first_seen": datetime.now(timezone.utc).strftime("%Y-%m-%d"),
-        "last_confirmed": datetime.now(timezone.utc).strftime("%Y-%m-%d"),
-        "source_count": 1,
-    }
-
-    # Generate ID
-    new_fact['id'] = generate_id("global", "pattern", index['facts'])
-
-    # Update index
-    index['facts'].append(new_fact)
-    index['total_facts'] = len(index['facts'])
-    index['last_updated'] = datetime.now(timezone.utc).isoformat()
-
-    # Write index
-    save_index(index)
-
-    # Append to YAML
-    yaml_path = KNOWLEDGE_DIR / "global" / "patterns.yaml"
-    yaml_path.parent.mkdir(parents=True, exist_ok=True)
-    mode = 'a' if yaml_path.exists() else 'w'
-    with open(yaml_path, mode, encoding='utf-8') as f:
-        if mode == 'w':
-            f.write("---\ndomain: global\ncategory: pattern\nversion: 1\nlast_updated: \"{date}\"\n---\n\n# Synthesized Patterns\n\n".format(date=datetime.now(timezone.utc).strftime("%Y-%m-%d")))
-        f.write(f"\n- id: {new_fact['id']}\n")
-        f.write(f"  fact: \"{hypothesis}\"\n")
-        f.write(f"  confidence: {plaus}\n")
-        if new_fact['tags']:
-            f.write(f"  tags: {json.dumps(new_fact['tags'])}\n")
-        f.write(f"  related: {json.dumps(source_ids)}\n")
-        f.write(f"  first_seen: \"{new_fact['first_seen']}\"\n")
-        f.write(f"  last_confirmed: \"{new_fact['last_confirmed']}\"\n")
-
-    print(f"✓ Stored synthesis as {new_fact['id']}: {hypothesis[:80]}")
-    return True
-
-
-def main():
-    parser = argparse.ArgumentParser(description="Zero-shot knowledge synthesis")
-    parser.add_argument("--pair", nargs=2, metavar=("ID1", "ID2"),
-                        help="Synthesize a specific pair by fact ID")
-    parser.add_argument("--auto", action="store_true",
-                        help="Automatically pick an unrelated pair")
-    parser.add_argument("--threshold", type=float, default=0.6,
-                        help="Plausibility threshold for storage (default: 0.6)")
-    parser.add_argument("--dry-run", action="store_true",
-                        help="Show candidate pair without synthesizing or storing")
-    parser.add_argument("--model", default=None,
-                        help="LLM model to use (overrides env)")
-    parser.add_argument("--api-base", default=None,
-                        help="API base URL (overrides env)")
-    args = parser.parse_args()
-
-    # Resolve API credentials
-    api_base = args.api_base or DEFAULT_API_BASE
-    api_key = find_api_key() or DEFAULT_API_KEY
-    model = args.model or DEFAULT_MODEL
-
-    if not args.dry_run and not args.pair and not args.auto:
-        print("ERROR: Must specify either --pair ID1 ID2 or --auto", file=sys.stderr)
-        parser.print_help()
-        sys.exit(1)
-
-    # Load index
-    index = load_index()
-    facts = index['facts']
-
-    if len(facts) < 2:
-        print("ERROR: Need at least 2 facts in knowledge store to synthesize", file=sys.stderr)
-        sys.exit(1)
-
-    # Select facts
-    f1, f2 = None, None
-    if args.pair:
-        id1, id2 = args.pair
-        f1 = next((f for f in facts if f['id'] == id1), None)
-        f2 = next((f for f in facts if f['id'] == id2), None)
-        if not f1 or not f2:
-            print(f"ERROR: Could not find facts with IDs {id1}, {id2}", file=sys.stderr)
-            sys.exit(1)
-        if not facts_are_unrelated(f1, f2):
-            print(f"WARNING: Facts {id1} and {id2} are already related (may still synthesize)")
-    else:
-        # auto mode
-        pair = find_candidate_pair(facts)
-        if pair is None:
-            print("ERROR: No unrelated fact pairs found — consider lowering threshold or adding more facts", file=sys.stderr)
-            sys.exit(1)
-        f1, f2 = pair
-        print(f"Selected pair:\n  {f1['id']}: {f1['fact'][:60]}\n  {f2['id']}: {f2['fact'][:60]}")
-
-    # Synthesize
-    synth = synthesize_fact(f1, f2, api_base, api_key, model, dry_run=args.dry_run)
-    if synth is None:
-        sys.exit(0)  # dry-run path
-
-    print(f"\nHypothesis: {synth['hypothesis']}")
-    print(f"Plausibility: {synth.get('plausibility', 0.0):.2f}")
-    print(f"Bridging concepts: {synth.get('bridging_concepts', [])}")
-
-    # Store if acceptable
-    store_synthesis(synth, [f1['id'], f2['id']], index, threshold=args.threshold)
-
-
-if __name__ == '__main__':
-    main()

scripts/test_dependency_freshness.py

@@ -0,0 +1,179 @@
+#!/usr/bin/env python3
+"""Tests for scripts/dependency_freshness.py — 9.7 Dependency Freshness."""
+
+import json
+import os
+import sys
+from unittest.mock import patch, MagicMock
+
+# Import target module
+sys.path.insert(0, os.path.dirname(__file__) or ".")
+import importlib.util
+spec = importlib.util.spec_from_file_location(
+    "dependency_freshness",
+    os.path.join(os.path.dirname(__file__) or ".", "dependency_freshness.py")
+)
+mod = importlib.util.module_from_spec(spec)
+spec.loader.exec_module(mod)
+
+parse_requirements = mod.parse_requirements
+get_major_version = mod.get_major_version
+is_more_than_two_majors_behind = mod.is_more_than_two_majors_behind
+analyze_dependencies = mod.analyze_dependencies
+
+
+def test_parse_requirements_simple():
+    """Parse a simple package line."""
+    import tempfile
+    with tempfile.NamedTemporaryFile(mode='w', suffix='.txt', delete=False) as f:
+        f.write("requests\n")
+        tmp = f.name
+    try:
+        pkgs = parse_requirements(tmp)
+        assert pkgs == ["requests"], f"got {pkgs}"
+        print("PASS: test_parse_requirements_simple")
+    finally:
+        os.unlink(tmp)
+
+
+def test_parse_requirements_with_specifiers():
+    """Parse lines with version specifiers."""
+    import tempfile
+    with tempfile.NamedTemporaryFile(mode='w', suffix='.txt', delete=False) as f:
+        f.write("pytest>=8,<9\n")
+        f.write("aiohttp>=3.8\n")
+        tmp = f.name
+    try:
+        pkgs = parse_requirements(tmp)
+        assert pkgs == ["pytest", "aiohttp"], f"got {pkgs}"
+        print("PASS: test_parse_requirements_with_specifiers")
+    finally:
+        os.unlink(tmp)
+
+
+def test_parse_requirements_ignores_comments_and_blanks():
+    """Comments and blank lines are skipped."""
+    import tempfile
+    with tempfile.NamedTemporaryFile(mode='w', suffix='.txt', delete=False) as f:
+        f.write("# This is a comment\n")
+        f.write("\n")
+        f.write("   \n")
+        f.write("numpy\n")
+        f.write("# another comment\n")
+        tmp = f.name
+    try:
+        pkgs = parse_requirements(tmp)
+        assert pkgs == ["numpy"], f"got {pkgs}"
+        print("PASS: test_parse_requirements_ignores_comments_and_blanks")
+    finally:
+        os.unlink(tmp)
+
+
+def test_get_major_version_normal():
+    """Extract major version from typical semantic strings."""
+    assert get_major_version("1.2.3") == 1
+    assert get_major_version("3.4.5") == 3
+    assert get_major_version("0.11.0") == 0
+    print("PASS: test_get_major_version_normal")
+
+
+def test_get_major_version_with_rc():
+    """Prerelease versions still yield major number."""
+    assert get_major_version("2.0.0rc1") == 2
+    assert get_major_version("1.0.0a1") == 1
+    print("PASS: test_get_major_version_with_rc")
+
+
+def test_is_more_than_two_majors_behind():
+    """Difference >2 triggers True; <=2 triggers False."""
+    assert is_more_than_two_majors_behind("1.2.3", "4.0.0") is True
+    assert is_more_than_two_majors_behind("3.9.0", "4.0.0") is False
+    assert is_more_than_two_majors_behind("2.1.0", "5.2.0") is True
+    assert is_more_than_two_majors_behind("8.0.0", "9.0.0") is False
+    assert is_more_than_two_majors_behind("4.0.0", "4.0.0") is False
+    print("PASS: test_is_more_than_two_majors_behind")
+
+
+def test_analyze_dependencies_very_outdated():
+    """Flag packages more than 2 major versions behind."""
+    required = ["pkg_a", "pkg_b"]
+    installed = {"pkg_a": "1.0.0", "pkg_b": "3.5.2"}
+    outdated = {
+        "pkg_a": {"installed": "1.0.0", "latest": "4.0.0"},
+        "pkg_b": {"installed": "3.5.2", "latest": "4.0.0"},
+    }
+    very_out, missing, outdated_ok = analyze_dependencies(required, installed, outdated)
+    assert len(very_out) == 1 and very_out[0]["package"] == "pkg_a"
+    assert len(missing) == 0
+    assert len(outdated_ok) == 1 and outdated_ok[0]["package"] == "pkg_b"
+    print("PASS: test_analyze_dependencies_very_outdated")
+
+
+def test_analyze_dependencies_missing():
+    """Detect packages not installed at all."""
+    required = ["pkg_a", "pkg_missing"]
+    installed = {"pkg_a": "2.0.0"}
+    outdated = {"pkg_a": {"installed": "2.0.0", "latest": "3.0.0"}}
+    very_out, missing, outdated_ok = analyze_dependencies(required, installed, outdated)
+    assert "pkg_missing" in missing
+    assert len(very_out) == 0
+    assert len(outdated_ok) == 1
+    print("PASS: test_analyze_dependencies_missing")
+
+
+def test_analyze_dependencies_up_to_date():
+    """Packages up-to-date are not flagged."""
+    required = ["pkg_good"]
+    installed = {"pkg_good": "5.0.0"}
+    outdated = {}
+    very_out, missing, outdated_ok = analyze_dependencies(required, installed, outdated)
+    assert len(very_out) == 0
+    assert len(missing) == 0
+    assert len(outdated_ok) == 0
+    print("PASS: test_analyze_dependencies_up_to_date")
+
+
+def test_generate_human_report_contains_very_outdated():
+    """Human report includes very outdated packages."""
+    very_out = [
+        {"package": "oldpkg", "installed": "1.0", "latest": "4.0", "major_diff": 3}
+    ]
+    missing = []
+    outdated_ok = []
+    report = mod.generate_human_report(very_out, missing, outdated_ok, "requirements.txt")
+    assert "oldpkg" in report
+    assert "Installed: 1.0" in report
+    assert "Latest:    4.0" in report
+    assert "Major diff: 3" in report
+    print("PASS: test_generate_human_report_contains_very_outdated")
+
+
+def test_generate_json_report_structure():
+    """JSON report contains required keys."""
+    very_out = [{"package": "oldpkg", "installed": "1.0", "latest": "4.0", "major_diff": 3}]
+    missing = ["missing_pkg"]
+    outdated_ok = []
+    report_json = mod.generate_json_report(very_out, missing, outdated_ok, "requirements.txt")
+    data = json.loads(report_json)
+    assert "summary" in data
+    assert data["summary"]["very_outdated_count"] == 1
+    assert data["summary"]["missing_count"] == 1
+    assert "very_outdated" in data
+    assert "missing" in data
+    print("PASS: test_generate_json_report_structure")
+
+
+if __name__ == '__main__':
+    print("Running dependency_freshness test suite...")
+    test_parse_requirements_simple()
+    test_parse_requirements_with_specifiers()
+    test_parse_requirements_ignores_comments_and_blanks()
+    test_get_major_version_normal()
+    test_get_major_version_with_rc()
+    test_is_more_than_two_majors_behind()
+    test_analyze_dependencies_very_outdated()
+    test_analyze_dependencies_missing()
+    test_analyze_dependencies_up_to_date()
+    test_generate_human_report_contains_very_outdated()
+    test_generate_json_report_structure()
+    print("ALL TESTS PASSED.")

scripts/test_knowledge_synthesizer.py

@@ -1,235 +0,0 @@
-#!/usr/bin/env python3
-"""
-Tests for knowledge_synthesizer.py — zero-shot knowledge synthesis pipeline.
-
-Run with: python3 scripts/test_knowledge_synthesizer.py
-Or via pytest: pytest scripts/test_knowledge_synthesizer.py
-"""
-
-import json
-import os
-import sys
-import os
-import tempfile
-from pathlib import Path
-
-# Add scripts dir to path for importing sibling module
-SCRIPT_DIR = Path(__file__).resolve().parent
-sys.path.insert(0, str(SCRIPT_DIR))
-
-import importlib.util
-spec = importlib.util.spec_from_file_location(
-    "ks", os.path.join(str(SCRIPT_DIR), "knowledge_synthesizer.py")
-)
-ks = importlib.util.module_from_spec(spec)
-spec.loader.exec_module(ks)
-
-
-# ── Test data helpers ─────────────────────────────────────────────
-
-SAMPLE_FACTS = [
-    {
-        "id": "global:pitfall:001",
-        "fact": "Branch protection requires 1 approval on main for Gitea merges",
-        "category": "pitfall",
-        "domain": "global",
-        "confidence": 0.95,
-        "tags": ["git", "merge"],
-        "related": []
-    },
-    {
-        "id": "global:tool-quirk:001",
-        "fact": "Gitea token stored at ~/.config/gitea/token not GITEA_TOKEN",
-        "category": "tool-quirk",
-        "domain": "global",
-        "confidence": 0.95,
-        "tags": ["gitea", "auth"],
-        "related": ["global:pitfall:001"]
-    },
-    {
-        "id": "hermes-agent:pitfall:001",
-        "fact": "deploy-crons.py leaves jobs in mixed model format",
-        "category": "pitfall",
-        "domain": "hermes-agent",
-        "confidence": 0.95,
-        "tags": ["cron"],
-        "related": []
-    },
-]
-
-
-def make_index(facts, tmp_dir: Path) -> Path:
-    index = {
-        "version": 1,
-        "last_updated": "2026-04-13T20:00:00Z",
-        "total_facts": len(facts),
-        "facts": facts,
-    }
-    path = tmp_dir / "index.json"
-    with open(path, "w") as f:
-        json.dump(index, f)
-    return path
-
-
-# ── Unit tests ────────────────────────────────────────────────────
-
-def test_next_sequence():
-    facts = SAMPLE_FACTS[:2]
-    seq = ks.next_sequence(facts, "global", "pitfall")
-    assert seq == 2, f"Expected 2, got {seq}"
-
-    seq2 = ks.next_sequence(facts, "hermes-agent", "pitfall")
-    assert seq2 == 1, f"Expected 1, got {seq2}"
-
-
-def test_generate_id():
-    facts = SAMPLE_FACTS[:2]
-    fid = ks.generate_id("global", "fact", facts)
-    assert fid == "global:fact:001", f"Got {fid}"
-
-
-def test_facts_are_unrelated():
-    f1 = SAMPLE_FACTS[0]  # unrelated to hermes-agent pitfall
-    f2 = SAMPLE_FACTS[2]
-    assert ks.facts_are_unrelated(f1, f2) is True
-
-    f3 = SAMPLE_FACTS[1]  # related to f1
-    assert ks.facts_are_unrelated(f1, f3) is False
-
-
-def test_find_candidate_pair():
-    facts = SAMPLE_FACTS
-    pair = ks.find_candidate_pair(facts)
-    assert pair is not None, "Should find an unrelated pair"
-    f1, f2 = pair
-    assert ks.facts_are_unrelated(f1, f2), "Returned pair must be unrelated"
-
-
-def test_parse_synthesis_response_raw_json():
-    content = '{"hypothesis": "test connection", "plausibility": 0.8, "bridging_concepts": ["x"], "suggested_tags": ["a"]}'
-    result = ks.parse_synthesis_response(content)
-    assert result is not None
-    assert result["hypothesis"] == "test connection"
-    assert result["plausibility"] == 0.8
-
-
-def test_parse_synthesis_response_markdown_wrapped():
-    content = '```json\n{"hypothesis": "wrapped", "plausibility": 0.5}\n```'
-    result = ks.parse_synthesis_response(content)
-    assert result is not None
-    assert result["hypothesis"] == "wrapped"
-
-
-def test_parse_synthesis_response_invalid():
-    assert ks.parse_synthesis_response("not json") is None
-    assert ks.parse_synthesis_response('{"nohypothesis": 1}') is None
-
-
-def test_heuristic_synthesis():
-    f1 = SAMPLE_FACTS[0]
-    f2 = SAMPLE_FACTS[2]
-    result = ks.heuristic_synthesis(f1, f2)
-    assert "hypothesis" in result
-    assert "plausibility" in result
-    assert result["plausibility"] == 0.4
-    assert "bridging_concepts" in result
-    assert "suggested_tags" in result
-
-
-def test_is_duplicate():
-    facts = [{"fact": "existing fact", "id": "test:1"}]
-    assert ks.is_duplicate("existing fact", facts) is True
-    assert ks.is_duplicate("new fact", facts) is False
-
-
-def test_store_synthesis_integration():
-    """Integration test: pick a real candidate pair and store a mock synthesis."""
-    with tempfile.TemporaryDirectory() as tmp:
-        tmp_path = Path(tmp)
-        # Create fake knowledge dir with index
-        kdir = tmp_path / "knowledge"
-        kdir.mkdir()
-        index = {
-            "version": 1,
-            "last_updated": "2026-04-13T20:00:00Z",
-            "total_facts": 3,
-            "facts": SAMPLE_FACTS
-        }
-        with open(kdir / "index.json", "w") as f:
-            json.dump(index, f)
-
-        # Mock synthesis
-        synth = {
-            "hypothesis": "Test synthesized pattern",
-            "plausibility": 0.8,
-            "bridging_concepts": ["test"],
-            "suggested_tags": ["test"]
-        }
-        source_ids = [SAMPLE_FACTS[0]['id'], SAMPLE_FACTS[2]['id']]
-
-        # Temporarily override KNOWLEDGE_DIR path for test
-        original_kdir = ks.KNOWLEDGE_DIR
-        ks.KNOWLEDGE_DIR = kdir
-        try:
-            stored = ks.store_synthesis(synth, source_ids, index, threshold=0.5)
-            assert stored is True
-            assert index['total_facts'] == 4
-            new_fact = index['facts'][-1]
-            assert new_fact['fact'] == "Test synthesized pattern"
-            assert new_fact['category'] == "pattern"
-            assert new_fact['domain'] == "global"
-            assert new_fact['related'] == source_ids
-            assert new_fact['id'].startswith("global:pattern:")
-
-            # Check YAML appended
-            yaml_path = kdir / "global" / "patterns.yaml"
-            assert yaml_path.exists()
-            content = yaml_path.read_text()
-            assert "Test synthesized pattern" in content
-        finally:
-            ks.KNOWLEDGE_DIR = original_kdir
-
-
-# ── Smoke test ────────────────────────────────────────────────────
-
-def test_smoke_synthesizer_info():
-    """Sanity check: script can at least load and report current knowledge state."""
-    index = ks.load_index()
-    total = index.get('total_facts', 0)
-    facts = index.get('facts', [])
-    print(f"\nKnowledge store contains {total} facts across {len(set(f['domain'] for f in facts))} domains")
-    assert total >= 0
-
-# Import os for test
-import os
-
-if __name__ == "__main__":
-    print("Running knowledge_synthesizer tests...\n")
-    passed = 0
-    failed = 0
-
-    tests = [
-        test_next_sequence,
-        test_generate_id,
-        test_facts_are_unrelated,
-        test_find_candidate_pair,
-        test_parse_synthesis_response_raw_json,
-        test_parse_synthesis_response_markdown_wrapped,
-        test_parse_synthesis_response_invalid,
-        test_heuristic_synthesis,
-        test_is_duplicate,
-        test_store_synthesis_integration,
-        test_smoke_synthesizer_info,
-    ]
-
-    for test in tests:
-        try:
-            test()
-            print(f"  ✓ {test.__name__}")
-            passed += 1
-        except Exception as e:
-            import traceback; traceback.print_exc(); print(f"  ✗ {test.__name__}: {e}")
-            failed += 1
-
-    print(f"\n{passed} passed, {failed} failed")
-    sys.exit(0 if failed == 0 else 1)

templates/synthesis-prompt.md

@@ -1,47 +0,0 @@
-# Knowledge Synthesis Prompt
-
-## System Prompt
-
-You are a knowledge synthesis engine. Given two facts, you generate a novel hypothesis
-that connects them in a way no human would typically link — a zero-shot creative leap.
-
-## Task
-
-FACT A:
-{fact_a}
-
-FACT B:
-{fact_b}
-
-Generate a single JSON object:
-
-{
-  "hypothesis": "one concise sentence linking the two facts as a new, testable insight",
-  "plausibility": 0.0-1.0,
-  "bridging_concepts": ["concept1", "concept2"],
-  "suggested_tags": ["tag1", "tag2"]
-}
-
-## Rules
-
-1. The hypothesis must be a logical consequence of combining both facts.
-2. DO NOT restate either fact — produce genuinely new insight.
-3. Plausibility should reflect confidence given only these two facts.
-4. If no meaningful connection exists, return {"hypothesis":"","plausibility":0.0}.
-5. Output ONLY valid JSON — no markdown, no explanation.
-
-## Examples
-
-Input facts:
-- "Gitea PR creation requires branch protection approval (1+) on main"
-- "Git push hangs on large repos (pack.windowMemory=100m)"
-
-Hypothesis output:
-{
-  "hypothesis": "Branch protection triggers checks that inflate pack size, causing git push to hang on large repos",
-  "plausibility": 0.65,
-  "bridging_concepts": ["git", "gitea", "branch-protection", "push"],
-  "suggested_tags": ["git", "gitea", "performance"]
-}
-
-Output ONLY the JSON object.