Round 96: cross-category eval_harness score floor library registry

Add assert_prompt_meets_eval_harness_floor DRY helper, refactor per-category
floor tests, and test_prompt_eval_harness_score_library_registry.py for
library-wide score-floor falsifiability. Sync governance (GLOSSARY step 28,
panel record, Decision Index). 652 pytest; ROUTE-001/002/003 100%.

Author: johnteee

README.md

@@ -21,7 +21,7 @@ Full library docs: [reflective-prompt-library/README.md](reflective-prompt-libra
 ## Governance
 
 - **Contributing:** [CONTRIBUTING.md](CONTRIBUTING.md) — quality gates, routing maintenance (R8–R12), `make all`
-- **Panel record:** [multi-agent-panel-consensus](reflective-prompt-library/plans/multi-agent-panel-consensus-2026-06-25.md) — six-lens Socratic consensus (Rounds 1–95)
+- **Panel record:** [multi-agent-panel-consensus](reflective-prompt-library/plans/multi-agent-panel-consensus-2026-06-25.md) — six-lens Socratic consensus (Rounds 1–96)
 - **Operator playbook:** [GLOSSARY.md](reflective-prompt-library/GLOSSARY.md) — Governance Maintenance Playbook
 
 The repository contains:

reflective-prompt-library/GLOSSARY.md

@@ -337,7 +337,7 @@ Curated top-of-cheatsheet summary of high-confusion routing traps (ROUTE-002 hol
 
 ## Governance Maintenance Playbook / 治理維護手冊
 
-Ongoing upkeep after panel close (Rounds 1–95). Not agent instructions — operator checklist.
+Ongoing upkeep after panel close (Rounds 1–96). Not agent instructions — operator checklist.
 
 **Operational test:** Before router tuning, add fresh ROUTE-002/003 holdout phrases; run `make all`; record decisions in `PROJECT_KNOWLEDGE.md` Decision Index when governance surface changes.
 
@@ -368,3 +368,4 @@ Ongoing upkeep after panel close (Rounds 1–95). Not agent instructions — ope
 25. When adding composable prompts or editing eval_harness contract preambles, keep `PROMPT_CONTRACT_HEADINGS` / `PROMPT_EVAL_MIN_SCORE` in `prompt_eval_helpers.py` and run `test_prompt_contract_library_registry.py` plus per-category `test_*_prompts_eval_harness.py` guards.
 26. When editing composable prompt Purpose preambles, keep `Primary workflow surface(s)` / Supporting-lens lines via `assert_primary_workflow_surface_preamble` in `prompt_eval_helpers.py`; update `SUPPORTING_LENS_PRIMARY_SURFACE_BY_CATEGORY` for exemptions; run `test_prompt_primary_workflow_surface_library_registry.py` plus per-category `test_*_prompts_eval_harness.py` guards.
 27. When editing category workflow skill coverage tuples, keep frozen `*_COVER_WORKFLOW_SKILLS` in `test_*_prompts_eval_harness.py` aligned with `assert_category_workflow_skill_coverage`; `01-thinking` stays exempt (consumer graph); run `test_workflow_skill_coverage_library_registry.py`.
+28. When editing eval_harness score floors, keep `PROMPT_EVAL_MIN_SCORE` in `prompt_eval_helpers.py` and use `assert_prompt_meets_eval_harness_floor` in per-category `test_*_prompts_eval_harness.py` guards; run `test_prompt_eval_harness_score_library_registry.py`.

reflective-prompt-library/PROJECT_KNOWLEDGE.md

@@ -72,6 +72,7 @@ deferred promotions are recurrence-gated — see [panel backlog](plans/multi-age
 
 ## Decision Index
 
+- 2026-06-25 Round 96 panel — cross-category eval_harness score floor library registry (`test_prompt_eval_harness_score_library_registry.py`, DRY `assert_prompt_meets_eval_harness_floor`) → [record](plans/multi-agent-panel-consensus-2026-06-25.md)
 - 2026-06-25 Round 85 panel — composable prompt Primary workflow surface preamble guards (`test_*_prompts_eval_harness.py`) + Supporting-lens exemption → [record](plans/multi-agent-panel-consensus-2026-06-25.md)
 - 2026-06-25 Round 94 panel — cross-category Primary workflow surface preamble library registry (`test_prompt_primary_workflow_surface_library_registry.py`, DRY `assert_primary_workflow_surface_preamble`) → [record](plans/multi-agent-panel-consensus-2026-06-25.md)
 - 2026-06-25 Round 95 panel — cross-category workflow skill coverage library registry (`test_workflow_skill_coverage_library_registry.py`, DRY `assert_category_workflow_skill_coverage`) → [record](plans/multi-agent-panel-consensus-2026-06-25.md)

reflective-prompt-library/README.md

@@ -30,7 +30,7 @@ Pick **Strictness L1–L6** first (`skills/reflective-dispatch/SKILL.md`, [GLOSS
 
 ## Governance Panel Record
 
-Multi-agent Socratic consensus on project goals and the nine skills (Rounds 1–95, options A–HA) is recorded in [plans/multi-agent-panel-consensus-2026-06-25.md](plans/multi-agent-panel-consensus-2026-06-25.md). Run `make all` before claiming routing or governance changes are verified.
+Multi-agent Socratic consensus on project goals and the nine skills (Rounds 1–96, options A–HF) is recorded in [plans/multi-agent-panel-consensus-2026-06-25.md](plans/multi-agent-panel-consensus-2026-06-25.md). Run `make all` before claiming routing or governance changes are verified.
 
 ## Directory Map
 

reflective-prompt-library/plans/QUALITY_GATES_SUMMARY.md

@@ -314,7 +314,7 @@ ROUTE-002 measures unseen phrasing separately from ROUTE-001. Round 7 (2026-06-2
 2. **ROUTE-001/002/003 in CI** — 128 + 102 + 53 paraphrases at 100% consistency (seeded fixtures); `validate_route_fixture.py` gates minimum coverage
 3. **Governance validators** — links, lint, governance metadata, PROJECT_KNOWLEDGE, benchmark fixture, skill examples
 4. **Harness policy docs** — CONTRIBUTING, AGENTS, SKILL_INSTALLATION, maintenance playbook
-5. **Doc anti-drift** — `test_routing_contract.py`, cheatsheet parity tests, `test_readme_governance.py`, `test_thinking_prompts_eval_harness.py`, `test_engineering_prompts_eval_harness.py`, `test_prompt_cross_links.py`, `test_core_prompts_eval_harness.py`, `test_human_review_library_registry.py`, `test_prompt_skill_links_library_registry.py`, `test_prompt_contract_library_registry.py`, `test_prompt_primary_workflow_surface_library_registry.py`, `test_workflow_skill_coverage_library_registry.py`, `test_agent_prompts_eval_harness.py`, `test_context_prompts_eval_harness.py`, `test_domain_prompts_eval_harness.py`, `test_repo_prompts_eval_harness.py`, `test_validate_governance.py`, `test_validate_links.py`, `test_lint_skills.py`, `test_skill_module_contract.py` (Escalation subsection + Trigger/Methods/Output/Never; 640+ pytest anti-drift suite in CI); reciprocal thinking-lens ↔ skill checks and `00-core` + composable `Primary workflow surface(s)` ↔ `*_SKILL_LINKS` parity in `test_prompt_cross_links.py` (including strict Primary workflow surfaces parity via `test_thinking_lens_primary_surfaces_match_consumer_graph`); Human Review + Escalation route-target guards in thinking/skill contract tests; composable `Primary workflow surface(s)` / Supporting-lens preamble guards and composable `## Human Review` preamble guards (route to `reflective-risk`) via `prompt_eval_helpers.assert_human_review_preamble` in `test_*_prompts_eval_harness.py`; frozen `*_HUMAN_REVIEW_REQUIRED` / `*_HUMAN_REVIEW_EXEMPT` set parity across all prompt categories (Round 90); library-wide contract heading registry (`PROMPT_CONTRACT_HEADINGS`, Round 93); workflow skill coverage registry (`*_COVER_WORKFLOW_SKILLS`, Round 95)
+5. **Doc anti-drift** — `test_routing_contract.py`, cheatsheet parity tests, `test_readme_governance.py`, `test_thinking_prompts_eval_harness.py`, `test_engineering_prompts_eval_harness.py`, `test_prompt_cross_links.py`, `test_core_prompts_eval_harness.py`, `test_human_review_library_registry.py`, `test_prompt_skill_links_library_registry.py`, `test_prompt_contract_library_registry.py`, `test_prompt_primary_workflow_surface_library_registry.py`, `test_workflow_skill_coverage_library_registry.py`, `test_prompt_eval_harness_score_library_registry.py`, `test_agent_prompts_eval_harness.py`, `test_context_prompts_eval_harness.py`, `test_domain_prompts_eval_harness.py`, `test_repo_prompts_eval_harness.py`, `test_validate_governance.py`, `test_validate_links.py`, `test_lint_skills.py`, `test_skill_module_contract.py` (Escalation subsection + Trigger/Methods/Output/Never; 650+ pytest anti-drift suite in CI); reciprocal thinking-lens ↔ skill checks and `00-core` + composable `Primary workflow surface(s)` ↔ `*_SKILL_LINKS` parity in `test_prompt_cross_links.py` (including strict Primary workflow surfaces parity via `test_thinking_lens_primary_surfaces_match_consumer_graph`); Human Review + Escalation route-target guards in thinking/skill contract tests; composable `Primary workflow surface(s)` / Supporting-lens preamble guards and composable `## Human Review` preamble guards (route to `reflective-risk`) via `prompt_eval_helpers.assert_human_review_preamble` in `test_*_prompts_eval_harness.py`; frozen `*_HUMAN_REVIEW_REQUIRED` / `*_HUMAN_REVIEW_EXEMPT` set parity across all prompt categories (Round 90); library-wide contract heading registry (`PROMPT_CONTRACT_HEADINGS`, Round 93); workflow skill coverage registry (`*_COVER_WORKFLOW_SKILLS`, Round 95); eval_harness score floor registry (`PROMPT_EVAL_MIN_SCORE`, Round 96)
 
 ### Ongoing maintenance (not blockers)
 
@@ -384,4 +384,4 @@ Phase 1 quality-gate tooling and documentation are **complete**. Routing consist
 - ✅ Benchmark fixture gate plus optional manual benchmark runs
 - ✅ Research-backed design decisions
 
-The project is positioned to grow sustainably with quality discipline built in from the start. **No open implementation blockers** remain from panel Rounds 1–95; work is recurrence-gated maintenance per playbook. The next measurable quality target is **holdout expansion before router tuning** and optional manual baseline-vs-skill benchmark runs — not shipping new core skills without promotion evidence.
+The project is positioned to grow sustainably with quality discipline built in from the start. **No open implementation blockers** remain from panel Rounds 1–96; work is recurrence-gated maintenance per playbook. The next measurable quality target is **holdout expansion before router tuning** and optional manual baseline-vs-skill benchmark runs — not shipping new core skills without promotion evidence.

reflective-prompt-library/plans/multi-agent-panel-consensus-2026-06-25.md

@@ -2829,4 +2829,55 @@ User directive (repeat): review prompts, plans, skills, and Socratic/critical-th
 
 ---
 
-**Resealed 2026-06-25** after **Round 95** (options GW–HA). Workflow skill coverage is now library-registry checked across all `00-core`–`06-repo` categories with shared `assert_category_workflow_skill_coverage` and frozen tuples per harness (`01-thinking` exempt via empty tuple). Holdout expansion remains recurrence-gated maintenance.
+## Round 96 — cross-category eval_harness score floor library registry (2026-06-25)
+
+**Options HB–HF** | Six-lens panel (Opus, Codex, Gemini, Composer, Sakana, GLM)
+
+### Round 96 options
+
+| ID | Proposal | Verdict |
+| --- | --- | --- |
+| HB | DRY `assert_prompt_meets_eval_harness_floor` in `prompt_eval_helpers.py` | **Agree** |
+| HC | `test_prompt_eval_harness_score_library_registry.py` — score floor registry + library-wide sweep | **Agree** |
+| HD | GLOSSARY playbook step 28 + governance sync | **Agree** |
+| HE | ROUTE holdout expansion | **Defer** |
+| HF | Router / tenth skill / benchmark CI | **Reject** |
+
+### Round 96 verdict table
+
+| ID | Option | Verdict | Action |
+| --- | --- | --- | --- |
+| HB | DRY eval_harness score floor helper | **Agree** | `assert_prompt_meets_eval_harness_floor` |
+| HC | Score floor library registry | **Agree** | `test_prompt_eval_harness_score_library_registry.py` |
+| HD | Playbook + docs | **Agree** | step 28; panel round 96 sync |
+| HE | Holdout expansion | **Defer** | maintenance |
+| HF | Router/tenth skill/benchmark CI | **Reject** | no change |
+
+### Socratic rationale (Round 96)
+
+- **Opus:** Rounds 91–95 closed HR, cross-link, contract, primary-surface, and workflow-coverage registries; per-category `meets_eval_harness_floor` guards remain duplicated with no library-wide falsifiability.
+- **Codex:** Centralizing `assert_prompt_meets_eval_harness_floor` prevents score-floor drift; registry sweep catches regressions across all 49 prompts in one place.
+- **Gemini:** Reject duplicating `EvalHarness.evaluate_file` logic in seven harness files — token/cost of maintenance, not runtime.
+- **Composer:** IDE adopters need one playbook step when bumping `PROMPT_EVAL_MIN_SCORE`; library registry mirrors R93 contract pattern.
+- **Sakana:** Score floor is orthogonal to thinking consumer graph — all seven categories including `01-thinking` belong in registry.
+- **GLM:** TW surface unchanged; score floor stays English-only harness policy.
+
+**All roles agree.**
+
+## Implemented Changes (Round 96)
+
+- `plans/tests/prompt_eval_helpers.py`: `assert_prompt_meets_eval_harness_floor`
+- `plans/tests/test_*_prompts_eval_harness.py`: DRY eval_harness score floor guards
+- `plans/tests/test_prompt_eval_harness_score_library_registry.py`: cross-category registry + library glob parity + library-wide sweep
+- `GLOSSARY.md`: playbook Rounds 1–96; step 28 for eval_harness score floor library registry
+- `QUALITY_GATES_SUMMARY.md`: score floor registry note; panel Rounds 1–96; 650+ pytest floor
+- `PROJECT_KNOWLEDGE.md`: Decision Index Round 96 entry
+- `README.md`, `reflective-prompt-library/README.md`, `test_readme_governance.py`: panel round 96 sync
+
+## Verification (Round 96)
+
+- `make all`: 652 pytest + ROUTE-001/002/003 100%
+
+---
+
+**Resealed 2026-06-25** after **Round 96** (options HB–HF). Eval_harness score floors are now library-registry checked across all `00-core`–`06-repo` categories with shared `assert_prompt_meets_eval_harness_floor` and per-category `MIN_SCORE` aliases to `PROMPT_EVAL_MIN_SCORE`. Holdout expansion remains recurrence-gated maintenance.

reflective-prompt-library/plans/tests/prompt_eval_helpers.py

@@ -128,3 +128,17 @@ def assert_category_workflow_skill_coverage(
     for skill in required_skills:
         assert skill in text, f"{category_label} should reference {skill}"
 
+def assert_prompt_meets_eval_harness_floor(
+    prompt_path: Path,
+    harness,
+    repo_root: str,
+    min_score: float = PROMPT_EVAL_MIN_SCORE,
+) -> None:
+    """Prompt must meet eval_harness score floor (default PROMPT_EVAL_MIN_SCORE)."""
+    rel = str(prompt_path.relative_to(repo_root))
+    result = harness.evaluate_file(rel)
+    assert result["score"] >= min_score, (
+        f"{prompt_path.name} eval_harness score {result['score']}% < {min_score}%: "
+        f"{[(c['id'], c['result']) for c in result['checks']]}"
+    )
+

reflective-prompt-library/plans/tests/test_agent_prompts_eval_harness.py

@@ -9,7 +9,7 @@
 sys.path.insert(0, str(Path(__file__).parent))
 
 from eval_harness import EvalHarness  # noqa: E402
-from prompt_eval_helpers import assert_category_workflow_skill_coverage, assert_human_review_preamble, assert_primary_workflow_surface_preamble, prompts_with_human_review, assert_human_review_required_matches_detection, assert_human_review_exempt_have_no_preamble_section, assert_human_review_sets_partition, PROMPT_CONTRACT_HEADINGS, PROMPT_EVAL_MIN_SCORE, assert_prompt_contract_headings  # noqa: E402
+from prompt_eval_helpers import assert_category_workflow_skill_coverage, assert_human_review_preamble, assert_primary_workflow_surface_preamble, prompts_with_human_review, assert_human_review_required_matches_detection, assert_human_review_exempt_have_no_preamble_section, assert_human_review_sets_partition, PROMPT_CONTRACT_HEADINGS, PROMPT_EVAL_MIN_SCORE, assert_prompt_contract_headings, assert_prompt_meets_eval_harness_floor  # noqa: E402
 
 REQUIRED_HEADINGS = PROMPT_CONTRACT_HEADINGS
 MIN_SCORE = PROMPT_EVAL_MIN_SCORE
@@ -55,12 +55,7 @@ def test_agent_prompt_has_contract_headings(prompt_path: Path):
 
 @pytest.mark.parametrize("prompt_path", AGENT_PROMPTS, ids=lambda p: p.name)
 def test_agent_prompt_meets_eval_harness_floor(prompt_path: Path, harness: EvalHarness):
-    rel = str(prompt_path.relative_to(REPO_ROOT))
-    result = harness.evaluate_file(rel)
-    assert result["score"] >= MIN_SCORE, (
-        f"{prompt_path.name} eval_harness score {result['score']}% < {MIN_SCORE}%: "
-        f"{[(c['id'], c['result']) for c in result['checks']]}"
-    )
+    assert_prompt_meets_eval_harness_floor(prompt_path, harness, REPO_ROOT, MIN_SCORE)
 
 def test_agent_prompts_reference_workflow_skills():
     for prompt_path in AGENT_PROMPTS:

reflective-prompt-library/plans/tests/test_context_prompts_eval_harness.py

@@ -9,7 +9,7 @@
 sys.path.insert(0, str(Path(__file__).parent))
 
 from eval_harness import EvalHarness  # noqa: E402
-from prompt_eval_helpers import assert_category_workflow_skill_coverage, assert_human_review_preamble, assert_primary_workflow_surface_preamble, prompts_with_human_review, assert_human_review_required_matches_detection, assert_human_review_exempt_have_no_preamble_section, assert_human_review_sets_partition, PROMPT_CONTRACT_HEADINGS, PROMPT_EVAL_MIN_SCORE, assert_prompt_contract_headings  # noqa: E402
+from prompt_eval_helpers import assert_category_workflow_skill_coverage, assert_human_review_preamble, assert_primary_workflow_surface_preamble, prompts_with_human_review, assert_human_review_required_matches_detection, assert_human_review_exempt_have_no_preamble_section, assert_human_review_sets_partition, PROMPT_CONTRACT_HEADINGS, PROMPT_EVAL_MIN_SCORE, assert_prompt_contract_headings, assert_prompt_meets_eval_harness_floor  # noqa: E402
 
 REQUIRED_HEADINGS = PROMPT_CONTRACT_HEADINGS
 MIN_SCORE = PROMPT_EVAL_MIN_SCORE
@@ -52,12 +52,7 @@ def test_context_prompt_has_contract_headings(prompt_path: Path):
 
 @pytest.mark.parametrize("prompt_path", CONTEXT_PROMPTS, ids=lambda p: p.name)
 def test_context_prompt_meets_eval_harness_floor(prompt_path: Path, harness: EvalHarness):
-    rel = str(prompt_path.relative_to(REPO_ROOT))
-    result = harness.evaluate_file(rel)
-    assert result["score"] >= MIN_SCORE, (
-        f"{prompt_path.name} eval_harness score {result['score']}% < {MIN_SCORE}%: "
-        f"{[(c['id'], c['result']) for c in result['checks']]}"
-    )
+    assert_prompt_meets_eval_harness_floor(prompt_path, harness, REPO_ROOT, MIN_SCORE)
 
 def test_context_prompts_reference_workflow_skills():
     for prompt_path in CONTEXT_PROMPTS:

reflective-prompt-library/plans/tests/test_core_prompts_eval_harness.py

@@ -13,7 +13,8 @@
     PROMPT_CONTRACT_HEADINGS,
     PROMPT_EVAL_MIN_SCORE,
     assert_primary_workflow_surface_preamble,
-    assert_category_workflow_skill_coverage, assert_prompt_contract_headings,  # noqa: E402
+    assert_category_workflow_skill_coverage, assert_prompt_contract_headings,
+    assert_prompt_meets_eval_harness_floor,  # noqa: E402
     assert_human_review_exempt_have_no_preamble_section,
     assert_human_review_preamble,
     assert_human_review_required_matches_detection,
@@ -61,12 +62,7 @@ def test_core_prompt_has_contract_headings(prompt_path: Path):
 
 @pytest.mark.parametrize("prompt_path", CORE_PROMPTS, ids=lambda p: p.name)
 def test_core_prompt_meets_eval_harness_floor(prompt_path: Path, harness: EvalHarness):
-    rel = str(prompt_path.relative_to(REPO_ROOT))
-    result = harness.evaluate_file(rel)
-    assert result["score"] >= MIN_SCORE, (
-        f"{prompt_path.name} eval_harness score {result['score']}% < {MIN_SCORE}%: "
-        f"{[(c['id'], c['result']) for c in result['checks']]}"
-    )
+    assert_prompt_meets_eval_harness_floor(prompt_path, harness, REPO_ROOT, MIN_SCORE)
 
 
 def test_core_prompts_reference_workflow_skills():