feat: Phase K K1 selection rationale schema -- auditable candidate selection (#812)

Makefile

@@ -1,4 +1,4 @@
-.PHONY: help demo test lint ci clean bench-leakage bench-multi-negatives bench-baseline bench-hidden-active bench-cluster-split bench-expert-ablation bench-expert-ablation-500 bench-selectivity bench-feature-decomp bench-gate bench-easy-baseline bench-charge-matched bench-order-dependent bench-precision-at-k bench-per-family bench-simulation-ablation bench-simulation-ablation-within-amp simulation-gate bench-strategy-compare regenerate-all generate phase3 pilot validate-scoring validate-scoring-phase3 validate-scoring-strict external-pilot pilot-confident presynth-qc gold-standard diversity synthesis-order novelty-broad external-consensus questionnaire gate-check ip-report benchmark-card wave0-5-gate-check wave0-5-novelty-audit wave0-5-novelty-audit-v2 wave0-5-panel wave0-5-evidence wave0-5-fill-external wave0-5b-generate wave0-5b-filter recalibration-engine recalibration-engine-dry-run validate-policy-version generate-synthetic-lab-results bump-policy-version generate-review-packet failed-candidate-report safe-publication-filter classify-negative-informativeness validate-rejection-events negative-result-dashboard calibration-audit calibration-audit-example calibration-overfit-check result-quality-filter synthetic-result-policy-check calibration-decision-checklist calibration-rollback-plan simulation-registry validate-simulation-result-schema simulation-baseline-check adapter-gate-check simulation-provenance simulation-ensemble-check simulation-ci-report simulation-deprecation-check simulation-scope-check simulation-evidence-packet artifact-version candidate-manifest artifact-changelog integration-check adapter-author-check license-check artifact-compat-check contribution-check adoption-scorecard release-gate-check decision-log release-request-check coi-check rotation-plan-check security-report-check citation-check roadmap-sync-check advisory-review-check annual-review-check
+.PHONY: help demo test lint ci clean bench-leakage bench-multi-negatives bench-baseline bench-hidden-active bench-cluster-split bench-expert-ablation bench-expert-ablation-500 bench-selectivity bench-feature-decomp bench-gate bench-easy-baseline bench-charge-matched bench-order-dependent bench-precision-at-k bench-per-family bench-simulation-ablation bench-simulation-ablation-within-amp simulation-gate bench-strategy-compare regenerate-all generate phase3 pilot validate-scoring validate-scoring-phase3 validate-scoring-strict external-pilot pilot-confident presynth-qc gold-standard diversity synthesis-order novelty-broad external-consensus questionnaire gate-check ip-report benchmark-card wave0-5-gate-check wave0-5-novelty-audit wave0-5-novelty-audit-v2 wave0-5-panel wave0-5-evidence wave0-5-fill-external wave0-5b-generate wave0-5b-filter recalibration-engine recalibration-engine-dry-run validate-policy-version generate-synthetic-lab-results bump-policy-version generate-review-packet failed-candidate-report safe-publication-filter classify-negative-informativeness validate-rejection-events negative-result-dashboard calibration-audit calibration-audit-example calibration-overfit-check result-quality-filter synthetic-result-policy-check calibration-decision-checklist calibration-rollback-plan simulation-registry validate-simulation-result-schema simulation-baseline-check adapter-gate-check simulation-provenance simulation-ensemble-check simulation-ci-report simulation-deprecation-check simulation-scope-check simulation-evidence-packet artifact-version candidate-manifest artifact-changelog integration-check adapter-author-check license-check artifact-compat-check contribution-check adoption-scorecard release-gate-check decision-log release-request-check coi-check rotation-plan-check security-report-check citation-check roadmap-sync-check advisory-review-check annual-review-check selection-rationale-check
 
 PYTHON := $(shell [ -f .venv/bin/python ] && echo .venv/bin/python || echo python3)
 PYTEST  := $(shell [ -f .venv/bin/pytest ] && echo .venv/bin/pytest || echo pytest)
@@ -885,3 +885,8 @@ annual-review-check:
 	.venv/bin/openamp-foundry annual-review-check \
 		--entry-json '{"review_id":"ANN-2026-001","year":"2026","section":"safety_policy","reviewer":"maintainer-1","finding_count":0,"action_items_count":0,"status":"completed","notes":"All safety checks passed.","completion_date":"2026-07-09"}'
 	@echo "Annual review check complete."
+
+selection-rationale-check:
+	.venv/bin/openamp-foundry selection-rationale-check \
+		--entry-json '{"selection_id":"SEL-2026-001","batch_id":"batch-2026-001","candidate_id":"OAMP-001","pipeline_version":"v0.7.9","selection_date":"2026-07-09","evidence_level":3,"baseline_comparison":"Scored above random baseline (AUROC 0.83) and charge-matched baseline (AUROC 0.71).","primary_criterion":"Top ensemble score with selectivity index > 2.","safety_flags_checked":["hemolysis_checked","toxicity_checked"],"reviewer":"maintainer-1"}'
+	@echo "Selection rationale check complete."

docs/evidence/METRICS_CURRENT.md

@@ -5,7 +5,9 @@ Machine-readable snapshot: `outputs/metrics_snapshot.json` regenerated with `mak
 > **Purpose:** One authoritative table of current pipeline metrics. If any doc disagrees
 > with this file, this file wins. Updated whenever benchmark/benchmark config changes.
 >
-> **Last updated:** 2026-07-09 (Phase J J10 — annual safety and benchmark review checklist — v0.7.8)
+> **Last updated:** 2026-07-09 (Phase K K1 — selection rationale schema — v0.7.9)
+> **New in v0.7.9:** Candidate selection rationale schema (K1 — Loop 119) — docs/evidence/SELECTION_RATIONALE_GUIDE.md with purpose, required field table (11 fields), evidence level table (6 levels from PROOF_LADDER.md), validation workflow. src/openamp_foundry/evidence/selection_rationale.py with SelectionRationaleEntry dataclass (11 fields: selection_id, batch_id, candidate_id, pipeline_version, selection_date, evidence_level, baseline_comparison, primary_criterion, safety_flags_checked, reviewer, dry_lab_only), SelectionRationaleResult dataclass (5 fields, dry_lab_only=True), VALID_EVIDENCE_LEVELS (6: 1-6), MINIMUM_SAFETY_FLAGS (1), validate_selection_rationale() (11 checks: SEL- prefix, non-empty batch_id/candidate_id/pipeline_version, YYYY-MM-DD date, valid evidence_level 1-6, non-empty baseline_comparison/primary_criterion, safety_flags_checked list with >=1 entry, non-empty reviewer, dry_lab_only must be True; evidence_level<=2 warns with synthesis caution), validate_selection_rationale_dict() (10 required fields guard). CLI (openamp-foundry selection-rationale-check) with --entry-json, --format text|json. make selection-rationale-check target. Phase K K1: every selection decision now has a machine-validated evidence trail.
+>
 > **New in v0.7.8:** Annual safety and benchmark review checklist (J10 — Loop 118) — docs/governance/ANNUAL_REVIEW_CHECKLIST.md with 5-section structured checklist (safety_policy: 6 checks including dual-use safeguards, dry_lab_only enforcement, toxicity filters; benchmark_thresholds: 6 checks including threshold loosening guard, easy-baseline requirement, deprecation check; calibration_status: 4 checks including recalibration gate and rollback plan; governance_decisions: 4 checks including COI disclosures and rotation plan; data_governance: 3 checks including proprietary data flag). src/openamp_foundry/governance/annual_review.py with AnnualReviewEntry dataclass (10 fields: review_id, year, section, reviewer, finding_count, action_items_count, status, notes, completion_date, dry_lab_only), AnnualReviewResult dataclass (6 fields: review_id, year, section, passed, errors, warnings, dry_lab_only=True), VALID_REVIEW_SECTIONS (5: benchmark_thresholds, calibration_status, data_governance, governance_decisions, safety_policy), VALID_ENTRY_STATUSES (5: completed, deferred, in_progress, not_applicable, pending), validate_annual_review_entry() (9 checks: ANN- prefix, 4-digit year, valid section, non-empty reviewer, non-negative finding_count, non-negative action_items_count, valid status, completed requires YYYY-MM-DD completion_date, dry_lab_only must be True; completed+no-notes warns, deferred warns, findings+no-action-items warns), validate_annual_review_dict() (7 required fields guard). CLI (openamp-foundry annual-review-check) with --entry-json, --format text|json. make annual-review-check target. Long-term trust: annual review entries are now machine-validated.
 >
 > **New in v0.7.7:** External advisory review process (J9 — Loop 117) — docs/governance/EXTERNAL_ADVISORY_REVIEW_PROCESS.md with reviewer eligibility criteria, review scope table (5 review types with minimum reviewer counts), 5-step review process (prepare packet, assign+disclose, receive+log, respond to findings, close), finding severity classification (critical/major/minor/informational), limitations. src/openamp_foundry/governance/advisory_review.py with AdvisoryReview dataclass (11 fields: review_id, review_type, artifact_id, reviewer_handle, assigned_date, deadline_date, status, finding_severity, finding_summary, resolved, dry_lab_only), AdvisoryReviewResult dataclass (5 fields: review_id, review_type, passed, errors, warnings, dry_lab_only=True), VALID_REVIEW_TYPES (5: benchmark_audit, candidate_review, evidence_review, governance_review, safety_policy_review), VALID_REVIEW_STATUSES (5: assigned, completed, deferred, in_progress, pending), VALID_FINDING_SEVERITIES (4: critical, informational, major, minor), MINIMUM_REVIEWER_COUNTS (5 entries: candidate_review and safety_policy_review require 2, others 1), validate_advisory_review() (9 checks: ADV- prefix, valid review_type, non-empty artifact_id, non-empty reviewer_handle, YYYY-MM-DD dates, valid status, valid finding_severity, dry_lab_only must be True; critical+unresolved warns, completed+no-summary warns, deferred warns), validate_advisory_review_dict() (7 required fields guard). CLI (openamp-foundry advisory-review-check) with --review-json, --format text|json. make advisory-review-check target. 29 tests. Credibility: external advisory reviews now have a validated structure and documented process.

docs/evidence/SELECTION_RATIONALE_GUIDE.md

@@ -0,0 +1,63 @@
+# Selection Rationale Guide
+
+> **Scope:** Dry-lab only. No wet-lab or in-vivo claims.
+> **Purpose:** Record WHY each candidate was selected for synthesis or further testing.
+> External reviewers and future calibration runs need this to audit and challenge decisions.
+
+---
+
+## What Is a Selection Rationale?
+
+A selection rationale (`SelectionRationaleEntry`) is a machine-validated record that
+documents, for each selected candidate:
+
+1. The evidence level (1–6 from `PROOF_LADDER.md`) supporting the selection.
+2. What the pipeline's score was compared to (baseline comparison).
+3. Which safety and quality flags were checked before selection.
+4. Who made the selection decision and when.
+
+## Required Fields
+
+| Field | Description |
+|---|---|
+| `selection_id` | Unique ID, must start with `SEL-` |
+| `batch_id` | The batch this selection belongs to |
+| `candidate_id` | The specific candidate being selected |
+| `pipeline_version` | Pipeline version used for scoring |
+| `selection_date` | Date of selection (YYYY-MM-DD) |
+| `evidence_level` | 1–6, must match `PROOF_LADDER.md` levels |
+| `baseline_comparison` | Description of what the score was compared to |
+| `primary_criterion` | Main reason this candidate was selected |
+| `safety_flags_checked` | List of safety/quality flags that were verified |
+| `reviewer` | Handle of the person or agent making the selection |
+| `dry_lab_only` | Must be True — no wet-lab claims |
+
+## Evidence Levels
+
+| Level | Meaning |
+|---|---|
+| 1 | Sequence-only computational nomination |
+| 2 | Multi-property computational nomination |
+| 3 | Computational nomination with novelty filter |
+| 4 | Computational nomination with calibration evidence |
+| 5 | Computational + simulation evidence (experimental still required) |
+| 6 | Experimental validation (requires human review before claim) |
+
+Levels 5 and 6 require `dry_lab_only=True` to avoid overclaiming.
+
+## How to Validate
+
+```bash
+make selection-rationale-check
+```
+
+or:
+
+```bash
+openamp-foundry selection-rationale-check \
+  --entry-json '{"selection_id": "SEL-2026-001", ...}'
+```
+
+---
+
+*This guide is machine-validated. See `src/openamp_foundry/evidence/selection_rationale.py`.*

docs/research/NEXT_100_PR_MAP.md

@@ -188,6 +188,19 @@ Make success survivable.
 | J9 | Add external advisory review process. | Credibility. | D |
 | J10 | Add annual safety and benchmark review checklist. | Long-term trust. | D |
 
+## Phase K — External Pilot Readiness
+
+Build the evidence and accountability artifacts needed for credible external validation.
+
+| PR | Task | Why it matters | Review class |
+|---:|---|---|---|
+| K1 | Add candidate selection rationale schema. | Auditable selection decisions. | B |
+| K2 | Add batch experiment priority ranker. | Rank candidates for next synthesis wave. | B |
+| K3 | Add pilot package completeness checker. | All components present before submission. | B |
+| K4 | Add post-experiment calibration intake schema. | Structured result-to-prediction comparison. | B/D |
+| K5 | Add uncertainty quantification report schema. | Honest prediction intervals for reviewers. | B |
+
+
 ## Prioritization rule
 
 When choosing between tasks, prefer the one that:

docs/research/ROADMAP.md

@@ -1,5 +1,20 @@
 # Roadmap
 
+## v0.7.9 — Loop 119: Phase K K1 — Selection Rationale Schema ✓ (2026-07-09)
+
+`docs/evidence/SELECTION_RATIONALE_GUIDE.md` documenting what selection rationale
+entries must contain, why they are needed for external review, and how evidence
+levels map to `PROOF_LADDER.md`.
+
+`src/openamp_foundry/evidence/selection_rationale.py` with `SelectionRationaleEntry`
+dataclass (11 fields, dry_lab_only=True enforced), `SelectionRationaleResult`
+dataclass (5 fields), `VALID_EVIDENCE_LEVELS` (1–6), `MINIMUM_SAFETY_FLAGS` (1),
+`validate_selection_rationale()` (11 checks, 1 warning condition for low evidence
+levels), `validate_selection_rationale_dict()` (10 required fields guard).
+CLI: `openamp-foundry selection-rationale-check`. `make selection-rationale-check` target.
+**Phase K milestone: v0.7.9** — every candidate selection now requires a
+machine-validated rationale with evidence level and baseline comparison.
+
 ## v0.7.8 — Loop 118: Phase J J10 — Annual Safety and Benchmark Review Checklist ✓ (2026-07-09)
 
 `docs/governance/ANNUAL_REVIEW_CHECKLIST.md` with 5-section structured annual

src/openamp_foundry/cli/commands/reports.py

@@ -2817,3 +2817,37 @@ def _run_annual_review_check(args: argparse.Namespace) -> int:
             print("  Annual review entry validated. Long-term trust maintained.")
 
     return 0 if result.passed else 3
+
+
+def _run_selection_rationale_check(args: argparse.Namespace) -> int:
+    """Validate a candidate selection rationale entry."""
+    import json as _json
+    from openamp_foundry.evidence.selection_rationale import validate_selection_rationale_dict
+
+    try:
+        d = _json.loads(args.entry_json)
+    except _json.JSONDecodeError as exc:
+        print(_json.dumps({"status": "error", "error": f"Invalid JSON: {exc}"}))
+        return 2
+
+    if not isinstance(d, dict):
+        print(_json.dumps({"status": "error", "error": "--entry-json must be a JSON object"}))
+        return 2
+
+    result = validate_selection_rationale_dict(d)
+    output_format = getattr(args, "format", "text")
+
+    if output_format == "json":
+        import dataclasses
+        print(_json.dumps(dataclasses.asdict(result), indent=2))
+    else:
+        status = "PASS" if result.passed else "FAIL"
+        print(f"Selection rationale {result.selection_id} (candidate {result.candidate_id}): {status}")
+        for e in result.errors:
+            print(f"  ERROR: {e}")
+        for w in result.warnings:
+            print(f"  WARN:  {w}")
+        if result.passed:
+            print("  Selection rationale validated. Decision is auditable.")
+
+    return 0 if result.passed else 3

src/openamp_foundry/cli/main.py

@@ -15,7 +15,7 @@
 from openamp_foundry.cli.commands.reports import _run_security_report_check
 from openamp_foundry.cli.commands.reports import _run_citation_check
 from openamp_foundry.cli.commands.reports import _run_roadmap_sync_check
-from openamp_foundry.cli.commands.reports import _run_advisory_review_check, _run_annual_review_check
+from openamp_foundry.cli.commands.reports import _run_advisory_review_check, _run_annual_review_check, _run_selection_rationale_check
 
 import argparse
 import json
@@ -1999,6 +1999,25 @@ def build_parser() -> argparse.ArgumentParser:
         help="Output format (default: text).",
     )
 
+    # ── Selection rationale check (Phase K K1) ───────────────────────
+    src2 = sub.add_parser(
+        "selection-rationale-check",
+        help=(
+            "Validate a candidate selection rationale. Ensures every selected "
+            "candidate has a documented evidence level and baseline comparison. "
+            "Dry-lab only."
+        ),
+    )
+    src2.add_argument(
+        "--entry-json", type=str, required=True,
+        help="JSON string of a SelectionRationaleEntry dict (required).",
+    )
+    src2.add_argument(
+        "--format", type=str, default="text",
+        choices=["text", "json"],
+        help="Output format (default: text).",
+    )
+
     # ── Annual review check (Phase J J10) ───────────────────────────
     anrc = sub.add_parser(
         "annual-review-check",
@@ -2285,6 +2304,9 @@ def main(argv: list[str] | None = None) -> int:
     if args.command == "annual-review-check":
         return _run_annual_review_check(args)
 
+    if args.command == "selection-rationale-check":
+        return _run_selection_rationale_check(args)
+
     parser.error("unknown command")
     return 2