diff --git a/docs/research/NEXT_100_PR_MAP.md b/docs/research/NEXT_100_PR_MAP.md
index 42ebbca4..80f9522d 100644
--- a/docs/research/NEXT_100_PR_MAP.md
+++ b/docs/research/NEXT_100_PR_MAP.md
@@ -118,7 +118,7 @@ Make failure useful.
 | F7 | Add calibration link from negative-result entries to intake reports (complete). — evidence/negative_result_calibration_link.py: links NRR- rejection records to calibration intake reports for closed-loop learning; tests/evidence/test_negative_result_calibration_link.py. | Closes learning loop. | C |
 | F8 | Add benchmark for whether rejected candidates resemble known failure modes (complete). — FMS- schema: 14 fields, 11 validation rules, PATTERN_REPEATED_THRESHOLD=0.80, pattern_repeated_flag enforcement, calibration_action_recommended; 63 tests. | Improves rejection logic. | C |
 | F9 | Add negative-result dashboard schema (complete). | NRD- aggregates NRR- rejection statistics: rejection rate consistency check, all_rejections_have_nrr enforced, top stage/reason controlled vocabulary, 100% rejection warning. | B |
-| F10 | Add policy that public claims must mention relevant negative results. | Prevents cherry-picking. | D |
+| F10 | Add policy that public claims must mention relevant negative results. | Prevents cherry-picking. | D | DONE |
 
 ## Phase G — Calibration and active-learning rigor
 
diff --git a/src/openamp_foundry/evidence/negative_result_citation_policy.py b/src/openamp_foundry/evidence/negative_result_citation_policy.py
new file mode 100644
index 00000000..9354eba7
--- /dev/null
+++ b/src/openamp_foundry/evidence/negative_result_citation_policy.py
@@ -0,0 +1,153 @@
+"""NRC- negative-result citation policy schema.
+
+Enforces that any public positive claim references all known relevant negative
+results.  Cherry-picking — highlighting successes while silently omitting
+failures — is the most common way an AI pipeline overstates its case.
+
+Every public claim about pipeline outputs must be paired with an NRC- record
+that lists relevant NRR- records and confirms they are cited.  A claim with
+uncited negatives is machine-flagged as non_compliant.
+"""
+
+from __future__ import annotations
+
+from dataclasses import dataclass
+
+VALID_NRC_VERDICTS: frozenset[str] = frozenset({
+    "compliant",
+    "non_compliant",
+    "no_relevant_negatives",
+})
+
+
+@dataclass
+class NegativeResultCitationPolicy:
+    nrc_id: str
+    claim_id: str
+    pipeline_version: str
+    relevant_nrr_ids: list[str]
+    cited_nrr_ids: list[str]
+    uncited_nrr_ids: list[str]
+    uncited_count: int
+    all_relevant_cited: bool
+    policy_verdict: str
+    dry_lab_only: bool
+    limitations: list[str]
+    created_at: str
+
+
+def validate_negative_result_citation_policy(nrc: NegativeResultCitationPolicy) -> None:
+    if not nrc.nrc_id.startswith("NRC-"):
+        raise ValueError(f"nrc_id must start with 'NRC-': {nrc.nrc_id!r}")
+    if not nrc.claim_id:
+        raise ValueError("claim_id must be non-empty")
+    if not nrc.pipeline_version:
+        raise ValueError("pipeline_version must be non-empty")
+    for nrr_id in nrc.relevant_nrr_ids:
+        if not nrr_id.startswith("NRR-"):
+            raise ValueError(f"relevant_nrr_id must start with 'NRR-': {nrr_id!r}")
+    for nrr_id in nrc.cited_nrr_ids:
+        if not nrr_id.startswith("NRR-"):
+            raise ValueError(f"cited_nrr_id must start with 'NRR-': {nrr_id!r}")
+    relevant_set = set(nrc.relevant_nrr_ids)
+    for nrr_id in nrc.cited_nrr_ids:
+        if nrr_id not in relevant_set:
+            raise ValueError(
+                f"cited_nrr_id {nrr_id!r} not in relevant_nrr_ids"
+            )
+    expected_uncited = [r for r in nrc.relevant_nrr_ids if r not in set(nrc.cited_nrr_ids)]
+    if sorted(nrc.uncited_nrr_ids) != sorted(expected_uncited):
+        raise ValueError("uncited_nrr_ids inconsistent with relevant_nrr_ids and cited_nrr_ids")
+    if nrc.uncited_count != len(nrc.uncited_nrr_ids):
+        raise ValueError("uncited_count must equal len(uncited_nrr_ids)")
+    expected_all_cited = nrc.uncited_count == 0
+    if nrc.all_relevant_cited != expected_all_cited:
+        raise ValueError("all_relevant_cited inconsistent with uncited_count")
+    if nrc.policy_verdict not in VALID_NRC_VERDICTS:
+        raise ValueError(
+            f"policy_verdict {nrc.policy_verdict!r} not in VALID_NRC_VERDICTS"
+        )
+    if not nrc.relevant_nrr_ids and nrc.policy_verdict != "no_relevant_negatives":
+        raise ValueError(
+            "policy_verdict must be 'no_relevant_negatives' when relevant_nrr_ids is empty"
+        )
+    if nrc.relevant_nrr_ids and nrc.all_relevant_cited and nrc.policy_verdict != "compliant":
+        raise ValueError(
+            "policy_verdict must be 'compliant' when all relevant negatives are cited"
+        )
+    if nrc.relevant_nrr_ids and not nrc.all_relevant_cited and nrc.policy_verdict != "non_compliant":
+        raise ValueError(
+            "policy_verdict must be 'non_compliant' when relevant negatives are uncited"
+        )
+    if not nrc.dry_lab_only:
+        raise ValueError("dry_lab_only must be True")
+    if not nrc.limitations:
+        raise ValueError("limitations must be non-empty")
+    if not nrc.created_at:
+        raise ValueError("created_at must be non-empty")
+
+
+def _compute_verdict(relevant: list[str], cited_set: set[str]) -> str:
+    if not relevant:
+        return "no_relevant_negatives"
+    if all(r in cited_set for r in relevant):
+        return "compliant"
+    return "non_compliant"
+
+
+def build_negative_result_citation_policy(
+    *,
+    nrc_id: str,
+    claim_id: str,
+    pipeline_version: str,
+    relevant_nrr_ids: list[str],
+    cited_nrr_ids: list[str],
+    limitations: list[str],
+    created_at: str,
+) -> NegativeResultCitationPolicy:
+    """Build a NegativeResultCitationPolicy.
+
+    uncited_nrr_ids, uncited_count, all_relevant_cited, and policy_verdict
+    are all auto-computed from relevant_nrr_ids and cited_nrr_ids.
+    """
+    relevant = list(relevant_nrr_ids)
+    cited = list(cited_nrr_ids)
+    cited_set = set(cited)
+    uncited = [r for r in relevant if r not in cited_set]
+    uncited_count = len(uncited)
+    all_cited = uncited_count == 0
+    verdict = _compute_verdict(relevant, cited_set)
+    nrc = NegativeResultCitationPolicy(
+        nrc_id=nrc_id,
+        claim_id=claim_id,
+        pipeline_version=pipeline_version,
+        relevant_nrr_ids=relevant,
+        cited_nrr_ids=cited,
+        uncited_nrr_ids=uncited,
+        uncited_count=uncited_count,
+        all_relevant_cited=all_cited,
+        policy_verdict=verdict,
+        dry_lab_only=True,
+        limitations=limitations,
+        created_at=created_at,
+    )
+    validate_negative_result_citation_policy(nrc)
+    return nrc
+
+
+def format_negative_result_citation_policy(nrc: NegativeResultCitationPolicy) -> str:
+    lines = [
+        f"Negative-Result Citation Policy — {nrc.nrc_id}",
+        f"Claim: {nrc.claim_id}  |  Pipeline: {nrc.pipeline_version}",
+        f"Verdict: {nrc.policy_verdict}",
+        f"Relevant negatives: {len(nrc.relevant_nrr_ids)}  "
+        f"Cited: {len(nrc.cited_nrr_ids)}  "
+        f"Uncited: {nrc.uncited_count}",
+    ]
+    if nrc.uncited_nrr_ids:
+        lines.append(f"Uncited negatives: {', '.join(nrc.uncited_nrr_ids)}")
+    lines.append(f"All relevant cited: {nrc.all_relevant_cited}")
+    lines.append(f"Limitations: {'; '.join(nrc.limitations)}")
+    lines.append(f"Created: {nrc.created_at}")
+    lines.append(f"dry_lab_only: {nrc.dry_lab_only}")
+    return "\n".join(lines)
diff --git a/tests/evidence/test_negative_result_citation_policy.py b/tests/evidence/test_negative_result_citation_policy.py
new file mode 100644
index 00000000..d085c863
--- /dev/null
+++ b/tests/evidence/test_negative_result_citation_policy.py
@@ -0,0 +1,314 @@
+"""Tests for NRC- negative-result citation policy schema."""
+
+import pytest
+from openamp_foundry.evidence.negative_result_citation_policy import (
+    NegativeResultCitationPolicy,
+    VALID_NRC_VERDICTS,
+    build_negative_result_citation_policy,
+    format_negative_result_citation_policy,
+    validate_negative_result_citation_policy,
+)
+
+# ---------------------------------------------------------------------------
+# Helpers
+# ---------------------------------------------------------------------------
+
+
+def _build(**kwargs):
+    defaults = dict(
+        nrc_id="NRC-001",
+        claim_id="CLM-001",
+        pipeline_version="v1.0",
+        relevant_nrr_ids=["NRR-001", "NRR-002"],
+        cited_nrr_ids=["NRR-001", "NRR-002"],
+        limitations=["dry-lab only"],
+        created_at="2026-07-10",
+    )
+    defaults.update(kwargs)
+    return build_negative_result_citation_policy(**defaults)
+
+
+# ---------------------------------------------------------------------------
+# 1. Constants
+# ---------------------------------------------------------------------------
+
+
+def test_valid_nrc_verdicts_is_frozenset():
+    assert isinstance(VALID_NRC_VERDICTS, frozenset)
+
+
+def test_valid_nrc_verdicts_contains_compliant():
+    assert "compliant" in VALID_NRC_VERDICTS
+
+
+def test_valid_nrc_verdicts_contains_non_compliant():
+    assert "non_compliant" in VALID_NRC_VERDICTS
+
+
+def test_valid_nrc_verdicts_contains_no_relevant_negatives():
+    assert "no_relevant_negatives" in VALID_NRC_VERDICTS
+
+
+def test_valid_nrc_verdicts_has_three_values():
+    assert len(VALID_NRC_VERDICTS) == 3
+
+
+# ---------------------------------------------------------------------------
+# 2. build happy paths
+# ---------------------------------------------------------------------------
+
+
+def test_build_returns_negative_result_citation_policy():
+    assert isinstance(_build(), NegativeResultCitationPolicy)
+
+
+def test_build_nrc_id_stored():
+    assert _build().nrc_id == "NRC-001"
+
+
+def test_build_claim_id_stored():
+    assert _build().claim_id == "CLM-001"
+
+
+def test_build_pipeline_version_stored():
+    assert _build().pipeline_version == "v1.0"
+
+
+def test_build_dry_lab_only_true():
+    assert _build().dry_lab_only is True
+
+
+def test_build_all_cited_gives_compliant():
+    r = _build(
+        relevant_nrr_ids=["NRR-001", "NRR-002"],
+        cited_nrr_ids=["NRR-001", "NRR-002"],
+    )
+    assert r.policy_verdict == "compliant"
+
+
+def test_build_no_relevant_gives_no_relevant_negatives():
+    r = _build(relevant_nrr_ids=[], cited_nrr_ids=[])
+    assert r.policy_verdict == "no_relevant_negatives"
+
+
+def test_build_uncited_gives_non_compliant():
+    r = _build(
+        relevant_nrr_ids=["NRR-001", "NRR-002"],
+        cited_nrr_ids=["NRR-001"],
+    )
+    assert r.policy_verdict == "non_compliant"
+
+
+def test_build_all_relevant_cited_true_when_compliant():
+    assert _build().all_relevant_cited is True
+
+
+def test_build_all_relevant_cited_false_when_non_compliant():
+    r = _build(
+        relevant_nrr_ids=["NRR-001", "NRR-002"],
+        cited_nrr_ids=["NRR-001"],
+    )
+    assert r.all_relevant_cited is False
+
+
+def test_build_uncited_count_zero_when_compliant():
+    assert _build().uncited_count == 0
+
+
+def test_build_uncited_count_correct_when_non_compliant():
+    r = _build(
+        relevant_nrr_ids=["NRR-001", "NRR-002"],
+        cited_nrr_ids=["NRR-001"],
+    )
+    assert r.uncited_count == 1
+
+
+def test_build_uncited_nrr_ids_correct():
+    r = _build(
+        relevant_nrr_ids=["NRR-001", "NRR-002"],
+        cited_nrr_ids=["NRR-001"],
+    )
+    assert r.uncited_nrr_ids == ["NRR-002"]
+
+
+def test_build_uncited_nrr_ids_empty_when_compliant():
+    assert _build().uncited_nrr_ids == []
+
+
+def test_build_no_relevant_uncited_count_zero():
+    r = _build(relevant_nrr_ids=[], cited_nrr_ids=[])
+    assert r.uncited_count == 0
+
+
+def test_build_relevant_nrr_ids_stored():
+    r = _build(relevant_nrr_ids=["NRR-001"], cited_nrr_ids=["NRR-001"])
+    assert "NRR-001" in r.relevant_nrr_ids
+
+
+def test_build_cited_nrr_ids_stored():
+    r = _build(relevant_nrr_ids=["NRR-001"], cited_nrr_ids=["NRR-001"])
+    assert "NRR-001" in r.cited_nrr_ids
+
+
+def test_build_limitations_stored():
+    assert _build().limitations == ["dry-lab only"]
+
+
+def test_build_created_at_stored():
+    assert _build().created_at == "2026-07-10"
+
+
+def test_build_single_relevant_single_cited_compliant():
+    r = _build(relevant_nrr_ids=["NRR-010"], cited_nrr_ids=["NRR-010"])
+    assert r.policy_verdict == "compliant"
+
+
+def test_build_many_relevant_none_cited_non_compliant():
+    relevant = [f"NRR-{i:03d}" for i in range(1, 6)]
+    r = _build(relevant_nrr_ids=relevant, cited_nrr_ids=[])
+    assert r.policy_verdict == "non_compliant"
+    assert r.uncited_count == 5
+
+
+def test_build_many_relevant_some_cited_non_compliant():
+    relevant = ["NRR-001", "NRR-002", "NRR-003"]
+    r = _build(relevant_nrr_ids=relevant, cited_nrr_ids=["NRR-001"])
+    assert r.uncited_count == 2
+
+
+# ---------------------------------------------------------------------------
+# 3. validate rejection cases
+# ---------------------------------------------------------------------------
+
+
+def test_validate_rejects_bad_nrc_id_prefix():
+    with pytest.raises(ValueError, match="NRC-"):
+        _build(nrc_id="BAD-001")
+
+
+def test_validate_rejects_empty_claim_id():
+    with pytest.raises(ValueError, match="claim_id"):
+        _build(claim_id="")
+
+
+def test_validate_rejects_empty_pipeline_version():
+    with pytest.raises(ValueError):
+        _build(pipeline_version="")
+
+
+def test_validate_rejects_bad_relevant_nrr_id_prefix():
+    with pytest.raises(ValueError, match="NRR-"):
+        _build(relevant_nrr_ids=["BAD-001"], cited_nrr_ids=[])
+
+
+def test_validate_rejects_bad_cited_nrr_id_prefix():
+    with pytest.raises(ValueError, match="NRR-"):
+        _build(
+            relevant_nrr_ids=["NRR-001"],
+            cited_nrr_ids=["BAD-001"],
+        )
+
+
+def test_validate_rejects_cited_not_in_relevant():
+    with pytest.raises(ValueError, match="not in relevant_nrr_ids"):
+        _build(
+            relevant_nrr_ids=["NRR-001"],
+            cited_nrr_ids=["NRR-999"],
+        )
+
+
+def test_validate_rejects_uncited_count_mismatch():
+    nrc = _build()
+    nrc.uncited_count = 99
+    with pytest.raises(ValueError, match="uncited_count"):
+        validate_negative_result_citation_policy(nrc)
+
+
+def test_validate_rejects_all_relevant_cited_mismatch():
+    nrc = _build()
+    nrc.all_relevant_cited = False
+    with pytest.raises(ValueError, match="all_relevant_cited"):
+        validate_negative_result_citation_policy(nrc)
+
+
+def test_validate_rejects_invalid_policy_verdict():
+    nrc = _build()
+    nrc.policy_verdict = "UNKNOWN"
+    with pytest.raises(ValueError, match="policy_verdict"):
+        validate_negative_result_citation_policy(nrc)
+
+
+def test_validate_rejects_compliant_when_no_relevant():
+    nrc = _build(relevant_nrr_ids=[], cited_nrr_ids=[])
+    nrc.policy_verdict = "compliant"
+    with pytest.raises(ValueError, match="no_relevant_negatives"):
+        validate_negative_result_citation_policy(nrc)
+
+
+def test_validate_rejects_non_compliant_when_all_cited():
+    nrc = _build()
+    nrc.policy_verdict = "non_compliant"
+    with pytest.raises(ValueError, match="compliant"):
+        validate_negative_result_citation_policy(nrc)
+
+
+def test_validate_rejects_dry_lab_only_false():
+    nrc = _build()
+    nrc.dry_lab_only = False
+    with pytest.raises(ValueError, match="dry_lab_only"):
+        validate_negative_result_citation_policy(nrc)
+
+
+def test_validate_rejects_empty_limitations():
+    with pytest.raises(ValueError, match="limitations"):
+        _build(limitations=[])
+
+
+def test_validate_rejects_empty_created_at():
+    with pytest.raises(ValueError):
+        _build(created_at="")
+
+
+# ---------------------------------------------------------------------------
+# 4. format
+# ---------------------------------------------------------------------------
+
+
+def test_format_contains_nrc_id():
+    assert "NRC-001" in format_negative_result_citation_policy(_build())
+
+
+def test_format_contains_claim_id():
+    assert "CLM-001" in format_negative_result_citation_policy(_build())
+
+
+def test_format_contains_pipeline_version():
+    assert "v1.0" in format_negative_result_citation_policy(_build())
+
+
+def test_format_contains_verdict():
+    assert "compliant" in format_negative_result_citation_policy(_build())
+
+
+def test_format_contains_uncited_nrr_when_non_compliant():
+    r = _build(
+        relevant_nrr_ids=["NRR-001", "NRR-002"],
+        cited_nrr_ids=["NRR-001"],
+    )
+    assert "NRR-002" in format_negative_result_citation_policy(r)
+
+
+def test_format_contains_all_relevant_cited():
+    assert "True" in format_negative_result_citation_policy(_build())
+
+
+def test_format_contains_limitations():
+    assert "dry-lab only" in format_negative_result_citation_policy(_build())
+
+
+def test_format_contains_dry_lab_only():
+    assert "dry_lab_only: True" in format_negative_result_citation_policy(_build())
+
+
+def test_format_is_string():
+    assert isinstance(format_negative_result_citation_policy(_build()), str)
diff --git a/tests/test_test_count_regression.py b/tests/test_test_count_regression.py
index dcbb0eee..95d4e4ce 100644
--- a/tests/test_test_count_regression.py
+++ b/tests/test_test_count_regression.py
@@ -4,7 +4,7 @@
 import sys
 import math
 
-BASELINE = 10924
+BASELINE = 10974
 
 
 def test_test_count_regression():