Add TDD strict mode and address eval-report code quality gaps.

Ship tdd_mode strict blocking with last_test_edits tracking, consolidate
path protection and finalize render helpers, add pytest.ini for stable
mechanical runs, and remove dead imports. Demo features skipped per request.

Co-authored-by: Cursor <cursoragent@cursor.com>

Author: jvalin17

gate/scripts/verify_gate.py

@@ -19,7 +19,6 @@ def _default_project_root() -> Path:
 
 from gate.attest import build_attestation, git_head_sha, write_attestation  # noqa: E402
 from gate.core import (  # noqa: E402
-    find_gates_config,
     load_gates_config,
     read_token,
     verify_token,

gates.json

@@ -8,6 +8,7 @@
   "auto": false,
   "continue": false,
   "tdd": true,
+  "tdd_mode": "remind",
   "skill_routing": true,
   "model": "auto",
   "gate_protect": true,

hooks/finalize_render.py

@@ -9,17 +9,44 @@
 from gate.attest import CheckResult
 
 
+def _utc_date() -> str:
+    return datetime.now(timezone.utc).strftime("%Y-%m-%d")
+
+
+def _report_header(skill: str, title: str, slug: str, report_id: str, extra_rows: str = "") -> str:
+    date = _utc_date()
+    return f"""<!-- agent-toolkit:{skill} | v1 | {date} | {report_id} -->
+<!-- writer: hooks/finalize_report.py — agent did not write this file -->
+# {title}
+
+| Field | Value |
+|-------|-------|
+| Status | completed |
+| Writer | hooks/finalize_report.py |
+| Skill | {skill} |
+| Slug | {slug} |
+| Date (UTC) | {date} |
+{extra_rows}"""
+
+
+def _append_summary_and_gate(md: str, summary: str, final_marker: str, pass_line: str = "") -> str:
+    if summary:
+        md += f"\n## Summary\n\n{summary}\n"
+    if pass_line:
+        md += f"\n{pass_line}\n"
+    md += f"\n## Final Gate\n\n{final_marker}\n"
+    return md
+
+
 def mechanical_table(test: CheckResult, lint: CheckResult) -> str:
     test_status = "passed" if test.passed else "FAILED"
     lint_status = "passed" if lint.passed else "FAILED"
-    test_detail = trim_detail(test.detail)
-    lint_detail = trim_detail(lint.detail)
     return f"""## Mechanical Re-run (hook-owned)
 
 | Check | Command | Result | Detail |
 |-------|---------|--------|--------|
-| tests | `{test.name}` | {test_status} | {test_detail} |
-| lint  | `{lint.name}` | {lint_status} | {lint_detail} |
+| tests | `{test.name}` | {test_status} | {trim_detail(test.detail)} |
+| lint  | `{lint.name}` | {lint_status} | {trim_detail(lint.detail)} |
 """
 
 
@@ -32,56 +59,29 @@ def compose_precommit_markdown(
     report_id: str,
 ) -> str:
     slug = findings["slug"]
-    date = datetime.now(timezone.utc).strftime("%Y-%m-%d")
     instr = findings["instructions"]
     rules = findings["rules"]
     readme = findings["readme"]
     tm = findings["tests_meaningful"]
     appv = findings["app_verification"]
     summary = findings.get("summary", "").strip()
 
-    test_status = "passed" if test.passed else "FAILED"
-    lint_status = "passed" if lint.passed else "FAILED"
-
     final_marker = (
         "[x] READY TO COMMIT\n[ ] BLOCKED"
         if ready
         else f"[ ] READY TO COMMIT\n[x] BLOCKED — {'; '.join(reasons)}"
     )
 
-    md = f"""<!-- agent-toolkit:precommit | v1 | {date} | {report_id} -->
-<!-- writer: hooks/finalize_report.py — agent did not write this file -->
-# Pre-commit Report: {slug}
-
-| Field | Value |
-|-------|-------|
-| Status | completed |
-| Writer | hooks/finalize_report.py |
-| Skill | precommit |
-| Slug | {slug} |
-| Date (UTC) | {date} |
-
-## Mechanical Re-run (hook-owned)
-
-| Check | Command | Result | Detail |
-|-------|---------|--------|--------|
-| tests | `{test.name}` | {test_status} | {trim_detail(test.detail)} |
-| lint  | `{lint.name}` | {lint_status} | {trim_detail(lint.detail)} |
-
-## Findings (agent-authored)
-
-- Instructions: {instr['addressed']}/{instr['total']} addressed
-- Test quality: {tm['result']}{inline(' — ', tm.get('evidence'))}
-- Rules: {rules['violations']} violation(s)
-- README: {'PASS' if readme['passed'] else 'FAIL'}{inline(' — ', readme.get('details'))}
-- App verification: {appv['status']}{inline(' — ', appv.get('notes'))}
-"""
-
-    if summary:
-        md += f"\n## Summary\n\n{summary}\n"
-
-    md += f"\n## Final Gate\n\n{final_marker}\n"
-    return md
+    md = _report_header("precommit", f"Pre-commit Report: {slug}", slug, report_id)
+    md += f"\n{mechanical_table(test, lint)}\n## Findings (agent-authored)\n\n"
+    md += (
+        f"- Instructions: {instr['addressed']}/{instr['total']} addressed\n"
+        f"- Test quality: {tm['result']}{inline(' — ', tm.get('evidence'))}\n"
+        f"- Rules: {rules['violations']} violation(s)\n"
+        f"- README: {'PASS' if readme['passed'] else 'FAIL'}{inline(' — ', readme.get('details'))}\n"
+        f"- App verification: {appv['status']}{inline(' — ', appv.get('notes'))}\n"
+    )
+    return _append_summary_and_gate(md, summary, final_marker)
 
 
 def compose_evaluate_markdown(
@@ -97,55 +97,26 @@ def compose_evaluate_markdown(
     slug = findings["slug"]
     topic = findings["topic"]
     dims = findings["dimensions"]
-    date = datetime.now(timezone.utc).strftime("%Y-%m-%d")
     grade = grade_letter(score)
     summary = findings.get("summary", "").strip()
 
-    dim_rows = []
-    for key, weight in EVAL_DIMENSION_WEIGHTS.items():
-        label = key.replace("_", " ").title()
-        dim_score = dims[key]
-        weighted = round(dim_score * weight)
-        dim_rows.append(
-            f"| {label} | {dim_score}% | {int(weight * 100)}% | {weighted} |"
-        )
-    dim_table = "\n".join(dim_rows)
-
-    if passed:
-        final_marker = (
-            f"[x] PASSED — score {score}% ≥ threshold {threshold}%\n"
-            f"[ ] BLOCKED"
-        )
-    else:
-        final_marker = f"[ ] PASSED\n[x] BLOCKED — {'; '.join(reasons)}"
-
-    md = f"""<!-- agent-toolkit:evaluate | v1 | {date} | {report_id} -->
-<!-- writer: hooks/finalize_report.py — agent did not write this file -->
-# Evaluation: {topic}
-# Score: **{score}%** ({grade})
-
-| Field | Value |
-|-------|-------|
-| Status | completed |
-| Writer | hooks/finalize_report.py |
-| Skill | evaluate |
-| Slug | {slug} |
-| Threshold | {threshold}% |
-| Date (UTC) | {date} |
-
-| Dimension | Score | Weight | Weighted |
-|-----------|-------|--------|----------|
-{dim_table}
-| **Overall** | | | **{score}%** |
-
-{mechanical_table(test, lint)}
-"""
+    dim_rows = [
+        f"| {key.replace('_', ' ').title()} | {dims[key]}% | {int(weight * 100)}% | {round(dims[key] * weight)} |"
+        for key, weight in EVAL_DIMENSION_WEIGHTS.items()
+    ]
 
-    if summary:
-        md += f"\n## Summary\n\n{summary}\n"
+    final_marker = (
+        f"[x] PASSED — score {score}% ≥ threshold {threshold}%\n[ ] BLOCKED"
+        if passed
+        else f"[ ] PASSED\n[x] BLOCKED — {'; '.join(reasons)}"
+    )
 
-    md += f"\n## Final Gate\n\n{final_marker}\n"
-    return md
+    extra = f"| Threshold | {threshold}% |\n"
+    md = _report_header("evaluate", f"Evaluation: {topic}\n# Score: **{score}%** ({grade})", slug, report_id, extra)
+    md += f"\n| Dimension | Score | Weight | Weighted |\n|-----------|-------|--------|----------|\n"
+    md += "\n".join(dim_rows)
+    md += f"\n| **Overall** | | | **{score}%** |\n\n{mechanical_table(test, lint)}\n"
+    return _append_summary_and_gate(md, summary, final_marker)
 
 
 def compose_reviewer_markdown(
@@ -159,9 +130,8 @@ def compose_reviewer_markdown(
     slug = findings["slug"]
     topic = findings["topic"]
     counts = findings["findings"]
-    date = datetime.now(timezone.utc).strftime("%Y-%m-%d")
     summary = findings.get("summary", "").strip()
-    areas = findings.get("areas_reviewed") or []
+    areas_line = ", ".join(findings.get("areas_reviewed") or []) or "—"
 
     if passed:
         final_marker = "[x] PASSED\n[ ] BLOCKED"
@@ -170,40 +140,14 @@ def compose_reviewer_markdown(
         final_marker = f"[ ] PASSED\n[x] BLOCKED — {'; '.join(reasons)}"
         pass_line = ""
 
-    areas_line = ", ".join(areas) if areas else "—"
-
-    md = f"""<!-- agent-toolkit:reviewer | v1 | {date} | {report_id} -->
-<!-- writer: hooks/finalize_report.py — agent did not write this file -->
-# Reviewer Report: {topic}
-
-| Field | Value |
-|-------|-------|
-| **Status** | completed |
-| Writer | hooks/finalize_report.py |
-| Skill | reviewer |
-| Slug | {slug} |
-| Areas reviewed | {areas_line} |
-| Date (UTC) | {date} |
-
-## Findings Summary
-
-| Severity | Count |
-|----------|-------|
-| High | {counts['high']} |
-| Medium | {counts['medium']} |
-| Low | {counts['low']} |
-
-{mechanical_table(test, lint)}
-"""
-
-    if summary:
-        md += f"\n## Summary\n\n{summary}\n"
-
-    if pass_line:
-        md += f"\n{pass_line}\n"
-
-    md += f"\n## Final Gate\n\n{final_marker}\n"
-    return md
+    extra = f"| Areas reviewed | {areas_line} |\n"
+    md = _report_header("reviewer", f"Reviewer Report: {topic}", slug, report_id, extra)
+    md += (
+        f"\n## Findings Summary\n\n| Severity | Count |\n|----------|-------|\n"
+        f"| High | {counts['high']} |\n| Medium | {counts['medium']} |\n| Low | {counts['low']} |\n\n"
+        f"{mechanical_table(test, lint)}\n"
+    )
+    return _append_summary_and_gate(md, summary, final_marker, pass_line)
 
 
 def compose_assess_markdown(
@@ -217,7 +161,6 @@ def compose_assess_markdown(
     slug = findings["slug"]
     topic = findings["topic"]
     counts = findings["findings"]
-    date = datetime.now(timezone.utc).strftime("%Y-%m-%d")
     summary = findings.get("summary", "").strip()
 
     if passed:
@@ -227,39 +170,12 @@ def compose_assess_markdown(
     else:
         final_marker = f"[ ] PASSED\n[x] BLOCKED — {'; '.join(reasons)}"
         pass_line = ""
-        fix_now_section = (
-            f"### [!!] Fix Now\n\n{counts['fix_now']} critical finding(s) remain.\n"
-        )
-
-    md = f"""<!-- agent-toolkit:assess | v1 | {date} | {report_id} -->
-<!-- writer: hooks/finalize_report.py — agent did not write this file -->
-# Assess Report: {topic}
-
-| Field | Value |
-|-------|-------|
-| **Status** | completed |
-| Writer | hooks/finalize_report.py |
-| Skill | assess |
-| Slug | {slug} |
-| Date (UTC) | {date} |
-
-## Findings Summary
-
-| Bucket | Count |
-|--------|-------|
-| [!!] Fix now | {counts['fix_now']} |
-| [~] Consider | {counts['consider']} |
-| [ok] Good as-is | {counts['good']} |
-
-{fix_now_section}
-{mechanical_table(test, lint)}
-"""
-
-    if summary:
-        md += f"\n## Summary\n\n{summary}\n"
-
-    if pass_line:
-        md += f"\n{pass_line}\n"
+        fix_now_section = f"### [!!] Fix Now\n\n{counts['fix_now']} critical finding(s) remain.\n"
 
-    md += f"\n## Final Gate\n\n{final_marker}\n"
-    return md
+    md = _report_header("assess", f"Assess Report: {topic}", slug, report_id)
+    md += (
+        f"\n## Findings Summary\n\n| Bucket | Count |\n|--------|-------|\n"
+        f"| [!!] Fix now | {counts['fix_now']} |\n| [~] Consider | {counts['consider']} |\n"
+        f"| [ok] Good as-is | {counts['good']} |\n\n{fix_now_section}{mechanical_table(test, lint)}\n"
+    )
+    return _append_summary_and_gate(md, summary, final_marker, pass_line)