Compare commits
1 Commits
step35/205
...
step35/173
| Author | SHA1 | Date | |
|---|---|---|---|
|
2f57c2b653 |
@@ -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()
|
||||
477
scripts/progress_tracker.py
Normal file
477
scripts/progress_tracker.py
Normal file
@@ -0,0 +1,477 @@
|
||||
#!/usr/bin/env python3
|
||||
"""
|
||||
Progress Tracker — Pipeline 10.8
|
||||
Track improvement metrics over time. Are we getting better?
|
||||
|
||||
Metrics tracked:
|
||||
1. Test coverage — % of Python functions with associated tests (test:source file ratio + line coverage if available)
|
||||
2. Doc coverage — % of Python callables with docstrings (AST-based)
|
||||
3. Issue close rate — closed / (opened + closed) per week (Gitea API)
|
||||
4. Dep freshness — % of requirements pinned vs outdated (pip list --outdated)
|
||||
|
||||
Output:
|
||||
- metrics/snapshots/YYYY-MM-DD.json — one snapshot per run
|
||||
- metrics/TRENDS.md — cumulative markdown table
|
||||
- stdout summary
|
||||
|
||||
Usage:
|
||||
python3 scripts/progress_tracker.py
|
||||
python3 scripts/progress_tracker.py --json
|
||||
python3 scripts/progress_tracker.py --output metrics/TRENDS.md
|
||||
|
||||
Weekly cron:
|
||||
0 9 * * 1 cd /path/to/compounding-intelligence && python3 scripts/progress_tracker.py
|
||||
"""
|
||||
|
||||
import argparse
|
||||
import json
|
||||
import os
|
||||
import re
|
||||
import subprocess
|
||||
import sys
|
||||
from collections import defaultdict
|
||||
from datetime import datetime, timezone, timedelta
|
||||
from pathlib import Path
|
||||
from typing import Any, Dict, List, Optional, Tuple
|
||||
|
||||
# ── Configuration ──────────────────────────────────────────────────────────
|
||||
|
||||
SCRIPT_DIR = Path(__file__).resolve().parent
|
||||
REPO_ROOT = SCRIPT_DIR.parent
|
||||
METRICS_DIR = REPO_ROOT / "metrics"
|
||||
SNAPSHOTS_DIR = METRICS_DIR / "snapshots"
|
||||
TOKEN_PATH = Path.home() / ".config" / "gitea" / "token"
|
||||
GITEA_API_BASE = "https://forge.alexanderwhitestone.com/api/v1"
|
||||
ORG = "Timmy_Foundation"
|
||||
|
||||
# Ensure paths exist
|
||||
SNAPSHOTS_DIR.mkdir(parents=True, exist_ok=True)
|
||||
|
||||
|
||||
# ── Helpers ─────────────────────────────────────────────────────────────────
|
||||
|
||||
def run_cmd(cmd: List[str], cwd: Path = REPO_ROOT) -> str:
|
||||
"""Run a shell command and return stdout (stderr merged)."""
|
||||
result = subprocess.run(
|
||||
cmd, capture_output=True, text=True, cwd=cwd, timeout=30
|
||||
)
|
||||
if result.returncode != 0:
|
||||
return ""
|
||||
return result.stdout.strip()
|
||||
|
||||
|
||||
def slugify_date(dt: datetime) -> str:
|
||||
return dt.strftime("%Y-%m-%d")
|
||||
|
||||
|
||||
def snapshot_path(dt: datetime) -> Path:
|
||||
return SNAPSHOTS_DIR / f"{slugify_date(dt)}.json"
|
||||
|
||||
|
||||
def load_snapshots() -> List[Dict[str, Any]]:
|
||||
"""Load all existing snapshots sorted by date."""
|
||||
snapshots = []
|
||||
for f in sorted(SNAPSHOTS_DIR.glob("*.json")):
|
||||
try:
|
||||
with open(f) as fp:
|
||||
snapshots.append(json.load(fp))
|
||||
except Exception:
|
||||
continue
|
||||
return snapshots
|
||||
|
||||
|
||||
# ── Metric 1: Test Coverage ─────────────────────────────────────────────────
|
||||
|
||||
def collect_test_coverage() -> Dict[str, Any]:
|
||||
"""
|
||||
Compute test coverage metrics.
|
||||
Counts test_*.py and *_test.py files vs non-test .py source files.
|
||||
Also attempts to read .coverage if present.
|
||||
"""
|
||||
all_py = list(REPO_ROOT.rglob("*.py"))
|
||||
|
||||
source_files = []
|
||||
test_files = []
|
||||
|
||||
for p in all_py:
|
||||
try:
|
||||
rel_parts = p.relative_to(REPO_ROOT).parts
|
||||
except ValueError:
|
||||
continue
|
||||
|
||||
# Skip hidden/cache/temp dirs (check only relative parts)
|
||||
if any(part.startswith('.') or part.startswith('__') for part in rel_parts):
|
||||
continue
|
||||
if any(part in ('node_modules', 'venv', '.venv', 'env', '.pytest_cache') for part in rel_parts):
|
||||
continue
|
||||
|
||||
if p.name.startswith("test_") or p.name.endswith("_test.py"):
|
||||
test_files.append(p)
|
||||
else:
|
||||
source_files.append(p)
|
||||
|
||||
# Try to get line coverage from .coverage
|
||||
coverage_percent = None
|
||||
coverage_tool = None
|
||||
coverage_file = REPO_ROOT / ".coverage"
|
||||
if coverage_file.exists():
|
||||
try:
|
||||
import coverage # type: ignore
|
||||
# Use coverage API if available
|
||||
cov = coverage.Coverage(data_file=str(coverage_file))
|
||||
cov.load()
|
||||
total = cov.report()
|
||||
coverage_percent = total if isinstance(total, float) else None
|
||||
coverage_tool = "coverage"
|
||||
except Exception:
|
||||
# Fallback: parse `coverage report` output
|
||||
out = run_cmd(["coverage", "report", "--skip-empty"])
|
||||
if out:
|
||||
for line in out.splitlines():
|
||||
if "TOTAL" in line:
|
||||
parts = line.split()
|
||||
if len(parts) >= 2:
|
||||
try:
|
||||
coverage_percent = float(parts[-1].rstrip('%'))
|
||||
coverage_tool = "coverage"
|
||||
break
|
||||
except ValueError:
|
||||
pass
|
||||
|
||||
return {
|
||||
"test_files": len(test_files),
|
||||
"source_files": len(source_files),
|
||||
"test_to_source_ratio": round(len(test_files) / len(source_files), 4) if source_files else 0.0,
|
||||
"coverage_tool": coverage_tool,
|
||||
"coverage_percent": coverage_percent,
|
||||
}
|
||||
|
||||
|
||||
# ── Metric 2: Doc Coverage ──────────────────────────────────────────────────
|
||||
|
||||
def collect_doc_coverage() -> Dict[str, Any]:
|
||||
"""
|
||||
Check AST of Python files for docstrings.
|
||||
Returns: callables_total, callables_with_doc, doc_coverage_percent
|
||||
"""
|
||||
import ast
|
||||
|
||||
all_py = list(REPO_ROOT.rglob("*.py"))
|
||||
|
||||
source_files = []
|
||||
test_files = []
|
||||
|
||||
for p in all_py:
|
||||
try:
|
||||
rel_parts = p.relative_to(REPO_ROOT).parts
|
||||
except ValueError:
|
||||
continue
|
||||
|
||||
if any(part.startswith('.') or part.startswith('__') for part in rel_parts):
|
||||
continue
|
||||
if any(part in ('node_modules', 'venv', '.venv', 'env', '.pytest_cache') for part in rel_parts):
|
||||
continue
|
||||
|
||||
if p.name.startswith("test_") or p.name.endswith("_test.py"):
|
||||
test_files.append(p)
|
||||
else:
|
||||
source_files.append(p)
|
||||
|
||||
total_callables = 0
|
||||
with_doc = 0
|
||||
|
||||
for p in source_files + test_files:
|
||||
try:
|
||||
with open(p) as f:
|
||||
tree = ast.parse(f.read(), filename=str(p))
|
||||
for node in ast.walk(tree):
|
||||
if isinstance(node, (ast.FunctionDef, ast.AsyncFunctionDef, ast.ClassDef)):
|
||||
total_callables += 1
|
||||
doc = ast.get_docstring(node)
|
||||
if doc and doc.strip():
|
||||
with_doc += 1
|
||||
except Exception:
|
||||
continue
|
||||
|
||||
return {
|
||||
"callables_total": total_callables,
|
||||
"callables_with_doc": with_doc,
|
||||
"doc_coverage_percent": round((with_doc / total_callables * 100) if total_callables else 0.0, 2),
|
||||
}
|
||||
|
||||
|
||||
# ── Metric 3: Issue Close Rate ──────────────────────────────────────────────
|
||||
|
||||
def collect_issue_metrics() -> Dict[str, Any]:
|
||||
"""
|
||||
Use Gitea API to get issue open/close stats for the last 7 days.
|
||||
Returns counts and close rate.
|
||||
"""
|
||||
token = ""
|
||||
if TOKEN_PATH.exists():
|
||||
token = TOKEN_PATH.read_text().strip()
|
||||
|
||||
if not token:
|
||||
return {
|
||||
"opened_last_7d": None,
|
||||
"closed_last_7d": None,
|
||||
"close_rate": None,
|
||||
"total_open": None,
|
||||
"note": "Gitea token not available"
|
||||
}
|
||||
|
||||
try:
|
||||
from urllib.request import Request, urlopen
|
||||
from urllib.error import HTTPError, URLError
|
||||
except ImportError:
|
||||
return {"error": "urllib not available"}
|
||||
|
||||
now = datetime.now(timezone.utc)
|
||||
week_ago = now - timedelta(days=7)
|
||||
since = week_ago.strftime("%Y-%m-%d")
|
||||
|
||||
headers = {"Authorization": f"token {token}"}
|
||||
base_url = f"{GITEA_API_BASE}/repos/{ORG}/compounding-intelligence/issues"
|
||||
|
||||
try:
|
||||
# Get issues from last 7 days
|
||||
url = f"{base_url}?state=all&since={since}&per_page=100"
|
||||
req = Request(url, headers=headers)
|
||||
with urlopen(req, timeout=15) as resp:
|
||||
issues = json.loads(resp.read())
|
||||
|
||||
opened = 0
|
||||
closed = 0
|
||||
for issue in issues:
|
||||
created = datetime.fromisoformat(issue["created_at"].replace("Z", "+00:00"))
|
||||
if created >= week_ago:
|
||||
opened += 1
|
||||
if issue.get("state") == "closed":
|
||||
closed_at_str = issue.get("closed_at")
|
||||
if closed_at_str:
|
||||
closed_at = datetime.fromisoformat(closed_at_str.replace("Z", "+00:00"))
|
||||
if closed_at >= week_ago:
|
||||
closed += 1
|
||||
|
||||
# Total open issues
|
||||
req2 = Request(f"{base_url}?state=open&per_page=1", headers=headers)
|
||||
with urlopen(req2, timeout=15) as resp:
|
||||
total_open = int(resp.headers.get("X-Total-Count", "0"))
|
||||
|
||||
total = opened + closed
|
||||
close_rate = closed / total if total > 0 else 0.0
|
||||
|
||||
return {
|
||||
"opened_last_7d": opened,
|
||||
"closed_last_7d": closed,
|
||||
"close_rate": round(close_rate, 4),
|
||||
"total_open": total_open,
|
||||
}
|
||||
except Exception as e:
|
||||
return {
|
||||
"opened_last_7d": None,
|
||||
"closed_last_7d": None,
|
||||
"close_rate": None,
|
||||
"total_open": None,
|
||||
"error": str(e)[:100],
|
||||
"note": "Gitea API unavailable"
|
||||
}
|
||||
|
||||
|
||||
# ── Metric 4: Dependency Freshness ─────────────────────────────────────────
|
||||
|
||||
def collect_dep_freshness() -> Dict[str, Any]:
|
||||
"""
|
||||
Check requirements.txt for outdated dependencies using pip list --outdated.
|
||||
Returns freshness percentage and outdated list.
|
||||
"""
|
||||
req_file = REPO_ROOT / "requirements.txt"
|
||||
if not req_file.exists():
|
||||
return {
|
||||
"total_deps": 0,
|
||||
"outdated_deps": 0,
|
||||
"freshness_percent": 100.0,
|
||||
"outdated_list": [],
|
||||
"note": "requirements.txt not found"
|
||||
}
|
||||
|
||||
# Parse requirements (very simple: take name before comparison op)
|
||||
reqs = []
|
||||
with open(req_file) as f:
|
||||
for line in f:
|
||||
line = line.strip()
|
||||
if not line or line.startswith("#"):
|
||||
continue
|
||||
m = re.match(r"^([a-zA-Z0-9_.-]+)", line)
|
||||
if m:
|
||||
reqs.append(m.group(1))
|
||||
|
||||
if not reqs:
|
||||
return {"total_deps": 0, "outdated_deps": 0, "freshness_percent": 100.0, "outdated_list": []}
|
||||
|
||||
# Query pip for outdated packages (may fail if pip not available)
|
||||
outdated_names = set()
|
||||
try:
|
||||
out = run_cmd(["pip", "list", "--outdated", "--format=json"])
|
||||
if out:
|
||||
data = json.loads(out)
|
||||
outdated_names = {item["name"].lower() for item in data}
|
||||
except Exception:
|
||||
pass
|
||||
|
||||
outdated = [p for p in reqs if p.lower() in outdated_names]
|
||||
total = len(reqs)
|
||||
outdated_count = len(outdated)
|
||||
freshness = round(((total - outdated_count) / total * 100) if total else 100.0, 1)
|
||||
|
||||
return {
|
||||
"total_deps": total,
|
||||
"outdated_deps": outdated_count,
|
||||
"freshness_percent": freshness,
|
||||
"outdated_list": outdated,
|
||||
}
|
||||
|
||||
|
||||
# ── Snapshot & Trends ───────────────────────────────────────────────────────
|
||||
|
||||
def take_snapshot() -> Dict[str, Any]:
|
||||
"""Collect all metrics and return a snapshot dict."""
|
||||
now = datetime.now(timezone.utc)
|
||||
test_cov = collect_test_coverage()
|
||||
doc_cov = collect_doc_coverage()
|
||||
issues = collect_issue_metrics()
|
||||
deps = collect_dep_freshness()
|
||||
|
||||
return {
|
||||
"timestamp": now.isoformat(),
|
||||
"date": slugify_date(now),
|
||||
"metrics": {
|
||||
"test_coverage": test_cov,
|
||||
"doc_coverage": doc_cov,
|
||||
"issues": issues,
|
||||
"dependencies": deps,
|
||||
}
|
||||
}
|
||||
|
||||
|
||||
def save_snapshot(snapshot: Dict[str, Any]) -> Path:
|
||||
path = snapshot_path(datetime.fromisoformat(snapshot["timestamp"]))
|
||||
with open(path, "w") as f:
|
||||
json.dump(snapshot, f, indent=2)
|
||||
return path
|
||||
|
||||
|
||||
def generate_trends(snapshots: List[Dict[str, Any]], output_path: Optional[Path] = None) -> str:
|
||||
"""Generate markdown trends table; optionally write to file."""
|
||||
if not snapshots:
|
||||
msg = "# Progress Tracker — Trends\n\nNo snapshots yet. Run `progress_tracker.py` to create the first snapshot."
|
||||
if output_path:
|
||||
output_path.parent.mkdir(parents=True, exist_ok=True)
|
||||
output_path.write_text(msg)
|
||||
return msg
|
||||
|
||||
lines = [
|
||||
"# Progress Tracker — Trends",
|
||||
f"\nLast updated: {datetime.now(timezone.utc).strftime('%Y-%m-%d %H:%M UTC')}",
|
||||
f"\nSnapshots: {len(snapshots)}\n",
|
||||
"| Date | Test Files → Source | Doc Coverage | Issues Closed/Opened (7d) | Dep Freshness |",
|
||||
"|------|---------------------|--------------|---------------------------|---------------|",
|
||||
]
|
||||
|
||||
for snap in reversed(snapshots): # chronological
|
||||
date = snap["date"]
|
||||
m = snap["metrics"]
|
||||
tc = m["test_coverage"]
|
||||
test_str = f"{tc['test_files']}/{tc['source_files']} ({tc['test_to_source_ratio']:.2f})"
|
||||
doc_str = f"{m['doc_coverage']['doc_coverage_percent']:.1f}%"
|
||||
issues_str = f"{m['issues'].get('closed_last_7d','-')}/{m['issues'].get('opened_last_7d','-')}"
|
||||
dep_str = f"{m['dependencies'].get('freshness_percent','?')}%"
|
||||
lines.append(f"| {date} | {test_str} | {doc_str} | {issues_str} | {dep_str} |")
|
||||
|
||||
# Current snapshot summary
|
||||
cur = snapshots[-1]
|
||||
cm = cur["metrics"]
|
||||
lines.append(f"\n## Current Snapshot ({cur['date']})\n")
|
||||
|
||||
tc = cm["test_coverage"]
|
||||
cov_line = f"- Test coverage: {tc['coverage_percent']:.1f}% (via {tc['coverage_tool']})\n" if tc["coverage_percent"] else "- Test coverage: (pytest-cov not configured)\n"
|
||||
lines.append(cov_line)
|
||||
lines.append(f"- Doc coverage: {cm['doc_coverage']['doc_coverage_percent']:.1f}%")
|
||||
|
||||
im = cm["issues"]
|
||||
if im.get("close_rate") is not None:
|
||||
lines.append(f"- Issue close rate (7d): {im['close_rate']*100:.1f}% ({im['closed_last_7d']} closed, {im['opened_last_7d']} opened)")
|
||||
else:
|
||||
lines.append(f"- Issue metrics: {im.get('note','unavailable')}")
|
||||
|
||||
dd = cm["dependencies"]
|
||||
lines.append(f"- Dep freshness: {dd.get('freshness_percent','?')}% outdated ({dd.get('outdated_deps',0)}/{dd.get('total_deps',0)} deps)")
|
||||
if dd.get('outdated_list'):
|
||||
lines.append(f" Outdated: {', '.join(dd['outdated_list'][:5])}")
|
||||
|
||||
content = "\n".join(lines) + "\n"
|
||||
|
||||
if output_path:
|
||||
output_path.parent.mkdir(parents=True, exist_ok=True)
|
||||
output_path.write_text(content)
|
||||
|
||||
return content
|
||||
|
||||
|
||||
# ── Main ─────────────────────────────────────────────────────────────────────
|
||||
|
||||
def main() -> int:
|
||||
parser = argparse.ArgumentParser(description="Progress Tracker — 10.8")
|
||||
parser.add_argument("--json", action="store_true", help="Emit snapshot as JSON only")
|
||||
parser.add_argument("--output", type=Path, default=METRICS_DIR / "TRENDS.md",
|
||||
help="Write trends markdown to this file")
|
||||
args = parser.parse_args()
|
||||
|
||||
snapshot = take_snapshot()
|
||||
all_snapshots = load_snapshots()
|
||||
path_written = save_snapshot(snapshot)
|
||||
|
||||
if args.json:
|
||||
print(json.dumps(snapshot, indent=2))
|
||||
return 0
|
||||
|
||||
trends = generate_trends(all_snapshots + [snapshot], output_path=args.output)
|
||||
|
||||
# Print current snapshot summary
|
||||
print(f"Snapshot saved: {path_written}\n")
|
||||
print(f"Progress Tracker — {snapshot['date']}")
|
||||
print("=" * 50)
|
||||
|
||||
m = snapshot["metrics"]
|
||||
tc = m["test_coverage"]
|
||||
print(f"Test files: {tc['test_files']} | Source files: {tc['source_files']} | Ratio: {tc['test_to_source_ratio']:.3f}")
|
||||
if tc["coverage_percent"] is not None:
|
||||
print(f"Line coverage: {tc['coverage_percent']:.1f}% (via {tc['coverage_tool']})")
|
||||
else:
|
||||
print("Line coverage: (not available — run `pytest --cov`)")
|
||||
|
||||
print()
|
||||
dc = m["doc_coverage"]
|
||||
print(f"Callables with docstrings: {dc['callables_with_doc']}/{dc['callables_total']} ({dc['doc_coverage_percent']:.1f}%)")
|
||||
|
||||
print()
|
||||
im = m["issues"]
|
||||
if im.get("close_rate") is not None:
|
||||
print(f"Issues (7d): {im['closed_last_7d']} closed / {im['opened_last_7d']} opened → close rate: {im['close_rate']*100:.1f}%")
|
||||
print(f"Total open: {im['total_open']}")
|
||||
else:
|
||||
print(f"Issues: {im.get('note','unavailable')}")
|
||||
|
||||
print()
|
||||
dd = m["dependencies"]
|
||||
print(f"Dependencies: {dd.get('total_deps',0)} total, {dd.get('outdated_deps',0)} outdated")
|
||||
if dd.get('outdated_list'):
|
||||
shown = dd['outdated_list'][:5]
|
||||
print(f"Outdated: {', '.join(shown)}" + ("..." if len(dd['outdated_list']) > 5 else ""))
|
||||
|
||||
print(f"\nTrends written to: {args.output}")
|
||||
return 0
|
||||
|
||||
|
||||
if __name__ == "__main__":
|
||||
sys.exit(main())
|
||||
@@ -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)
|
||||
@@ -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.
|
||||
Reference in New Issue
Block a user