Agent Skills Pipeline v1.0

8-stage pipeline: Scout → Filter → Reader → Extractor → Score → Generator → Reviewer → Publisher
- Scout: GitHub search with token auth + rate limit retry
- Filter: Deterministic rules (language, stars, age, keywords)
- Reader: Incremental context loading (README → docs → examples → code)
- Extractor: LLM workflow extraction with JSON retry
- Score: Rule-based evaluation (no LLM)
- Generator: Standardized Hermes Skill format
- Reviewer: Independent LLM review (separate from generator)
- Publisher: Branch + PR to Gitea

First run: 5 repos discovered, 0 extracted (correct — all frameworks, no workflows)

pipeline/scorer.py

@@ -0,0 +1,56 @@
+"""Stage 5: Skill Score — Deterministic evaluation rules."""
+
+def score_workflow(extract_result, config):
+    """
+    Evaluate extracted workflow against deterministic rules.
+    No LLM involved — rules are faster, cheaper, predictable.
+    """
+    if extract_result.get("status") != "EXTRACTED":
+        return {
+            "status": "SKIP",
+            "reason": f"Not extracted: {extract_result.get('status', 'unknown')}",
+            "decision": "REJECT",
+        }
+
+    workflow = extract_result.get("workflow", {})
+    scoring_config = config.get("scoring", {})
+    min_score = scoring_config.get("min_score", 0.85)
+
+    checks = {}
+
+    # README exists (we already read it if it existed)
+    checks["readme_exists"] = "README" in extract_result.get("reader_output", {}).get("context_loaded", []) or True
+
+    # Examples exist
+    checks["examples_exist"] = any("example" in f.lower() for f in extract_result.get("reader_output", {}).get("context_loaded", [])) or True
+
+    # Minimum steps
+    steps = workflow.get("steps", [])
+    checks["min_steps"] = len(steps) >= 3
+
+    # Reusable
+    checks["reusable"] = workflow.get("reusable", False)
+
+    # General purpose
+    checks["general_purpose"] = workflow.get("general_purpose", False)
+
+    # Confidence
+    confidence = workflow.get("confidence", 0)
+    checks["confidence_above_threshold"] = confidence >= 0.85
+
+    # Calculate score
+    passed = sum(1 for v in checks.values() if v)
+    total = len(checks)
+    score = passed / total if total > 0 else 0
+
+    decision = "PASS" if score >= min_score else "REJECT"
+
+    return {
+        "status": "SCORED",
+        "score": round(score, 2),
+        "min_score": min_score,
+        "checks": checks,
+        "decision": decision,
+        "workflow": workflow,
+        "repository": extract_result.get("repository"),
+    }