diff --git a/src/openamp_foundry/evidence/example_external_review_packet.py b/src/openamp_foundry/evidence/example_external_review_packet.py
new file mode 100644
index 00000000..e28eafdb
--- /dev/null
+++ b/src/openamp_foundry/evidence/example_external_review_packet.py
@@ -0,0 +1,173 @@
+"""
+Example external review packet using toy/mock data.
+
+E2: Partners know what to expect. Provides a factory function that builds
+a fully-validated ExternalReviewPacket with synthetic candidates only.
+Candidate IDs use the TOY- prefix to prevent confusion with real candidates.
+This module is CI-checkable: build_example_packet() must pass all validation rules.
+"""
+
+from __future__ import annotations
+
+from dataclasses import dataclass, field
+
+from openamp_foundry.evidence.external_review_packet import (
+    BenchmarkSummary,
+    CalibrationSummary,
+    CandidateEntry,
+    ExternalReviewPacket,
+    PacketValidationResult,
+    SafetyAttestations,
+    validate_external_review_packet,
+)
+
+TOY_PACKET_ID: str = "ERP-TOY-0001"
+TOY_VERSION: str = "0.1.0"
+TOY_PIPELINE_VERSION: str = "openamp-foundry==0.1.0-example"
+TOY_GIT_SHA: str = "0000000000000000000000000000000000000000"
+TOY_CONTACT: str = "example@openamp-foundry.example"
+
+VALID_TOY_CANDIDATE_ID_PREFIX: str = "TOY-"
+
+
+@dataclass
+class ToyCandidate:
+    """Structured toy candidate for E2 example packet."""
+
+    candidate_id: str
+    sequence: str
+    ensemble_score: float
+    family: str
+    selection_rationale: str
+    safety_notes: str
+
+
+TOY_CANDIDATES: list[ToyCandidate] = [
+    ToyCandidate(
+        candidate_id="TOY-0001",
+        sequence="GLLDFLSLAAKFAAADYL",
+        ensemble_score=0.82,
+        family="alpha-helical",
+        selection_rationale="High ensemble score; representative of alpha-helical family.",
+        safety_notes="Toy candidate only. Not screened. Not for synthesis.",
+    ),
+    ToyCandidate(
+        candidate_id="TOY-0002",
+        sequence="KWKLFKKIGAVLKVL",
+        ensemble_score=0.76,
+        family="cationic",
+        selection_rationale="Cationic family representative with moderate ensemble score.",
+        safety_notes="Toy candidate only. Not screened. Not for synthesis.",
+    ),
+    ToyCandidate(
+        candidate_id="TOY-0003",
+        sequence="RRWWRWWRR",
+        ensemble_score=0.71,
+        family="tryptophan-rich",
+        selection_rationale="Tryptophan-rich motif; included for family diversity.",
+        safety_notes="Toy candidate only. Not screened. Not for synthesis.",
+    ),
+]
+
+
+def build_toy_candidate_entry(toy: ToyCandidate) -> CandidateEntry:
+    """Convert a ToyCandidate into a CandidateEntry suitable for an ExternalReviewPacket."""
+    if not toy.candidate_id.startswith(VALID_TOY_CANDIDATE_ID_PREFIX):
+        raise ValueError(
+            f"Toy candidate IDs must start with '{VALID_TOY_CANDIDATE_ID_PREFIX}', "
+            f"got: {toy.candidate_id!r}"
+        )
+    return CandidateEntry(
+        candidate_id=toy.candidate_id,
+        sequence=toy.sequence,
+        ensemble_score=toy.ensemble_score,
+        proof_ladder_level=1,
+        family=toy.family,
+        selection_rationale=toy.selection_rationale,
+        safety_notes=toy.safety_notes,
+    )
+
+
+def build_example_benchmark_summary() -> BenchmarkSummary:
+    """Return a toy BenchmarkSummary for the E2 example packet."""
+    return BenchmarkSummary(
+        auroc=0.75,
+        benchmark_name="BMC-EXAMPLE-toy-precision-at-k",
+        n_positives=50,
+        n_negatives=150,
+    )
+
+
+def build_example_calibration_summary() -> CalibrationSummary:
+    """Return a toy CalibrationSummary for the E2 example packet."""
+    return CalibrationSummary(
+        calibration_assessment="moderately_calibrated",
+        n_calibration_samples=100,
+    )
+
+
+def build_example_safety_attestations() -> SafetyAttestations:
+    """Return SafetyAttestations appropriate for a toy example packet."""
+    return SafetyAttestations(
+        reviewed_by_human=False,
+        safety_gate_passed=True,
+        no_known_toxicity_claim=True,
+        hemolysis_assay_mandatory=True,
+    )
+
+
+def build_example_packet(
+    packet_id: str = TOY_PACKET_ID,
+    version: str = TOY_VERSION,
+    generated_at: str = "2026-01-01T00:00:00Z",
+    pipeline_version: str = TOY_PIPELINE_VERSION,
+    git_sha: str = TOY_GIT_SHA,
+    contact: str = TOY_CONTACT,
+) -> ExternalReviewPacket:
+    """
+    Build an ExternalReviewPacket populated with toy/mock candidates only.
+
+    This packet is designed to pass all validation rules and serve as a
+    reference example for external reviewers. Candidate IDs use the TOY-
+    prefix; the packet is marked dry_lab_only_attestation=True.
+    """
+    candidates = [build_toy_candidate_entry(toy) for toy in TOY_CANDIDATES]
+    return ExternalReviewPacket(
+        packet_id=packet_id,
+        version=version,
+        generated_at=generated_at,
+        pipeline_version=pipeline_version,
+        git_sha=git_sha,
+        candidate_count=len(candidates),
+        candidates=candidates,
+        benchmark_summary=build_example_benchmark_summary(),
+        calibration_summary=build_example_calibration_summary(),
+        limitations=[
+            "All candidates are synthetic toy data. No biological screening performed.",
+            "AUROC is illustrative only; not computed on real AMP datasets.",
+            "Calibration assessment is illustrative; not computed on real assay outcomes.",
+        ],
+        safety_attestations=build_example_safety_attestations(),
+        dry_lab_only_attestation=True,
+        proof_ladder_level=1,
+        contact=contact,
+        notes="E2 example packet. Contains toy candidates only. Not for clinical or research use.",
+    )
+
+
+def validate_example_packet_passes() -> PacketValidationResult:
+    """
+    Build and validate the example packet.
+
+    Returns a PacketValidationResult with is_valid=True if the example
+    packet satisfies all ExternalReviewPacket validation rules.
+    Raises ValueError if validation fails (which would indicate a regression
+    in either the example data or the validation logic).
+    """
+    packet = build_example_packet()
+    result = validate_external_review_packet(packet)
+    if not result.is_valid:
+        raise ValueError(
+            f"E2 example packet failed validation: {result.violations}"
+        )
+    return result
diff --git a/tests/evidence/test_example_external_review_packet.py b/tests/evidence/test_example_external_review_packet.py
new file mode 100644
index 00000000..6f80d239
--- /dev/null
+++ b/tests/evidence/test_example_external_review_packet.py
@@ -0,0 +1,313 @@
+"""Tests for E2 example external review packet (toy/mock data only)."""
+
+from __future__ import annotations
+
+import pytest
+
+from openamp_foundry.evidence.example_external_review_packet import (
+    TOY_CANDIDATES,
+    TOY_CONTACT,
+    TOY_GIT_SHA,
+    TOY_PACKET_ID,
+    TOY_PIPELINE_VERSION,
+    TOY_VERSION,
+    VALID_TOY_CANDIDATE_ID_PREFIX,
+    ToyCandidate,
+    build_example_benchmark_summary,
+    build_example_calibration_summary,
+    build_example_packet,
+    build_example_safety_attestations,
+    build_toy_candidate_entry,
+    validate_example_packet_passes,
+)
+from openamp_foundry.evidence.external_review_packet import (
+    validate_external_review_packet,
+)
+
+
+class TestToyConstants:
+    def test_toy_packet_id_prefix(self):
+        assert TOY_PACKET_ID.startswith("ERP-")
+
+    def test_toy_packet_id_contains_toy(self):
+        assert "TOY" in TOY_PACKET_ID
+
+    def test_toy_version_is_string(self):
+        assert isinstance(TOY_VERSION, str)
+        assert len(TOY_VERSION) > 0
+
+    def test_toy_pipeline_version_is_string(self):
+        assert isinstance(TOY_PIPELINE_VERSION, str)
+        assert "example" in TOY_PIPELINE_VERSION.lower()
+
+    def test_toy_git_sha_length(self):
+        assert len(TOY_GIT_SHA) == 40
+
+    def test_toy_git_sha_is_zeros(self):
+        assert TOY_GIT_SHA == "0" * 40
+
+    def test_toy_contact_is_string(self):
+        assert isinstance(TOY_CONTACT, str)
+        assert "@" in TOY_CONTACT
+
+    def test_valid_toy_prefix(self):
+        assert VALID_TOY_CANDIDATE_ID_PREFIX == "TOY-"
+
+
+class TestToyCandidates:
+    def test_three_toy_candidates(self):
+        assert len(TOY_CANDIDATES) == 3
+
+    def test_all_ids_have_toy_prefix(self):
+        for toy in TOY_CANDIDATES:
+            assert toy.candidate_id.startswith("TOY-"), toy.candidate_id
+
+    def test_all_sequences_nonempty(self):
+        for toy in TOY_CANDIDATES:
+            assert len(toy.sequence) > 0
+
+    def test_all_scores_in_range(self):
+        for toy in TOY_CANDIDATES:
+            assert 0.0 <= toy.ensemble_score <= 1.0
+
+    def test_all_families_nonempty(self):
+        for toy in TOY_CANDIDATES:
+            assert len(toy.family) > 0
+
+    def test_all_rationales_nonempty(self):
+        for toy in TOY_CANDIDATES:
+            assert len(toy.selection_rationale) > 0
+
+    def test_all_safety_notes_mention_toy(self):
+        for toy in TOY_CANDIDATES:
+            assert "toy" in toy.safety_notes.lower() or "not for" in toy.safety_notes.lower()
+
+    def test_toy_candidate_is_dataclass(self):
+        toy = TOY_CANDIDATES[0]
+        assert isinstance(toy, ToyCandidate)
+
+    def test_distinct_candidate_ids(self):
+        ids = [t.candidate_id for t in TOY_CANDIDATES]
+        assert len(ids) == len(set(ids))
+
+    def test_distinct_sequences(self):
+        seqs = [t.sequence for t in TOY_CANDIDATES]
+        assert len(seqs) == len(set(seqs))
+
+
+class TestBuildToyCandidate:
+    def test_build_returns_candidate_entry(self):
+        from openamp_foundry.evidence.external_review_packet import CandidateEntry
+        entry = build_toy_candidate_entry(TOY_CANDIDATES[0])
+        assert isinstance(entry, CandidateEntry)
+
+    def test_candidate_id_preserved(self):
+        entry = build_toy_candidate_entry(TOY_CANDIDATES[0])
+        assert entry.candidate_id == TOY_CANDIDATES[0].candidate_id
+
+    def test_sequence_preserved(self):
+        entry = build_toy_candidate_entry(TOY_CANDIDATES[0])
+        assert entry.sequence == TOY_CANDIDATES[0].sequence
+
+    def test_score_preserved(self):
+        entry = build_toy_candidate_entry(TOY_CANDIDATES[0])
+        assert entry.ensemble_score == TOY_CANDIDATES[0].ensemble_score
+
+    def test_proof_ladder_level_is_1(self):
+        entry = build_toy_candidate_entry(TOY_CANDIDATES[0])
+        assert entry.proof_ladder_level == 1
+
+    def test_family_preserved(self):
+        entry = build_toy_candidate_entry(TOY_CANDIDATES[0])
+        assert entry.family == TOY_CANDIDATES[0].family
+
+    def test_rationale_preserved(self):
+        entry = build_toy_candidate_entry(TOY_CANDIDATES[0])
+        assert entry.selection_rationale == TOY_CANDIDATES[0].selection_rationale
+
+    def test_safety_notes_preserved(self):
+        entry = build_toy_candidate_entry(TOY_CANDIDATES[0])
+        assert entry.safety_notes == TOY_CANDIDATES[0].safety_notes
+
+    def test_invalid_prefix_raises(self):
+        bad = ToyCandidate(
+            candidate_id="REAL-0001",
+            sequence="ACDEF",
+            ensemble_score=0.5,
+            family="x",
+            selection_rationale="x",
+            safety_notes="x",
+        )
+        with pytest.raises(ValueError, match="TOY-"):
+            build_toy_candidate_entry(bad)
+
+    def test_empty_prefix_raises(self):
+        bad = ToyCandidate(
+            candidate_id="0001",
+            sequence="ACDEF",
+            ensemble_score=0.5,
+            family="x",
+            selection_rationale="x",
+            safety_notes="x",
+        )
+        with pytest.raises(ValueError):
+            build_toy_candidate_entry(bad)
+
+
+class TestBuildExampleHelpers:
+    def test_benchmark_summary_auroc_in_range(self):
+        bm = build_example_benchmark_summary()
+        assert 0.0 <= bm.auroc <= 1.0
+
+    def test_benchmark_summary_name_nonempty(self):
+        bm = build_example_benchmark_summary()
+        assert len(bm.benchmark_name) > 0
+
+    def test_benchmark_summary_positives_positive(self):
+        bm = build_example_benchmark_summary()
+        assert bm.n_positives > 0
+
+    def test_benchmark_summary_negatives_positive(self):
+        bm = build_example_benchmark_summary()
+        assert bm.n_negatives > 0
+
+    def test_calibration_summary_valid_assessment(self):
+        from openamp_foundry.evidence.external_review_packet import VALID_CALIBRATION_ASSESSMENTS
+        cs = build_example_calibration_summary()
+        assert cs.calibration_assessment in VALID_CALIBRATION_ASSESSMENTS
+
+    def test_calibration_summary_samples_positive(self):
+        cs = build_example_calibration_summary()
+        assert cs.n_calibration_samples > 0
+
+    def test_safety_attestations_no_toxicity_claim(self):
+        sa = build_example_safety_attestations()
+        assert sa.no_known_toxicity_claim is True
+
+    def test_safety_attestations_hemolysis_mandatory(self):
+        sa = build_example_safety_attestations()
+        assert sa.hemolysis_assay_mandatory is True
+
+    def test_safety_attestations_gate_passed(self):
+        sa = build_example_safety_attestations()
+        assert sa.safety_gate_passed is True
+
+
+class TestBuildExamplePacket:
+    def setup_method(self):
+        self.packet = build_example_packet()
+
+    def test_packet_id_is_default(self):
+        assert self.packet.packet_id == TOY_PACKET_ID
+
+    def test_packet_id_has_erp_prefix(self):
+        assert self.packet.packet_id.startswith("ERP-")
+
+    def test_dry_lab_attestation_true(self):
+        assert self.packet.dry_lab_only_attestation is True
+
+    def test_candidate_count_matches_list(self):
+        assert self.packet.candidate_count == len(self.packet.candidates)
+
+    def test_candidate_count_is_three(self):
+        assert self.packet.candidate_count == 3
+
+    def test_all_candidate_ids_toy_prefix(self):
+        for c in self.packet.candidates:
+            assert c.candidate_id.startswith("TOY-")
+
+    def test_proof_ladder_level_at_most_2(self):
+        from openamp_foundry.evidence.external_review_packet import DRY_LAB_MAX_PROOF_LADDER_LEVEL
+        assert self.packet.proof_ladder_level <= DRY_LAB_MAX_PROOF_LADDER_LEVEL
+
+    def test_limitations_nonempty(self):
+        assert len(self.packet.limitations) > 0
+
+    def test_limitations_mention_toy(self):
+        combined = " ".join(self.packet.limitations).lower()
+        assert "toy" in combined or "synthetic" in combined
+
+    def test_contact_is_toy_contact(self):
+        assert self.packet.contact == TOY_CONTACT
+
+    def test_notes_mention_example(self):
+        assert "example" in self.packet.notes.lower()
+
+    def test_notes_mention_not_for_use(self):
+        assert "not for" in self.packet.notes.lower()
+
+    def test_version_is_default(self):
+        assert self.packet.version == TOY_VERSION
+
+    def test_pipeline_version_is_default(self):
+        assert self.packet.pipeline_version == TOY_PIPELINE_VERSION
+
+    def test_git_sha_is_default(self):
+        assert self.packet.git_sha == TOY_GIT_SHA
+
+    def test_custom_packet_id(self):
+        packet = build_example_packet(packet_id="ERP-TOY-9999")
+        assert packet.packet_id == "ERP-TOY-9999"
+
+    def test_custom_generated_at(self):
+        packet = build_example_packet(generated_at="2030-01-01T00:00:00Z")
+        assert packet.generated_at == "2030-01-01T00:00:00Z"
+
+    def test_custom_contact(self):
+        packet = build_example_packet(contact="reviewer@lab.example")
+        assert packet.contact == "reviewer@lab.example"
+
+
+class TestValidateExamplePacket:
+    def test_example_packet_is_valid(self):
+        result = validate_example_packet_passes()
+        assert result.is_valid is True
+
+    def test_example_packet_no_violations(self):
+        result = validate_example_packet_passes()
+        assert result.violations == []
+
+    def test_example_packet_has_packet_id(self):
+        result = validate_example_packet_passes()
+        assert result.packet_id == TOY_PACKET_ID
+
+    def test_validate_directly(self):
+        packet = build_example_packet()
+        result = validate_external_review_packet(packet)
+        assert result.is_valid is True
+
+    def test_validate_summary_nonempty(self):
+        result = validate_example_packet_passes()
+        assert len(result.validation_summary) > 0
+
+    def test_invalid_proof_ladder_fails(self):
+        from openamp_foundry.evidence.external_review_packet import DRY_LAB_MAX_PROOF_LADDER_LEVEL
+        packet = build_example_packet()
+        packet.proof_ladder_level = DRY_LAB_MAX_PROOF_LADDER_LEVEL + 1
+        result = validate_external_review_packet(packet)
+        assert not result.is_valid
+
+    def test_empty_limitations_fails(self):
+        packet = build_example_packet()
+        packet.limitations = []
+        result = validate_external_review_packet(packet)
+        assert not result.is_valid
+
+    def test_false_dry_lab_attestation_fails(self):
+        packet = build_example_packet()
+        packet.dry_lab_only_attestation = False
+        result = validate_external_review_packet(packet)
+        assert not result.is_valid
+
+    def test_mismatched_candidate_count_fails(self):
+        packet = build_example_packet()
+        packet.candidate_count = 999
+        result = validate_external_review_packet(packet)
+        assert not result.is_valid
+
+    def test_empty_candidates_fails(self):
+        packet = build_example_packet()
+        packet.candidates = []
+        packet.candidate_count = 0
+        result = validate_external_review_packet(packet)
+        assert not result.is_valid