diff --git a/case/evidence_index.sample.json b/case/evidence_index.sample.json
new file mode 100644
index 0000000..4022d2e
--- /dev/null
+++ b/case/evidence_index.sample.json
@@ -0,0 +1,59 @@
+{
+  "case_id": "my_real_case",
+  "generated_at": "2025-11-21T00:00:00Z",
+  "notes": "Sample evidence index. Replace with real export from validator.",
+  "evidence": [
+    {
+      "id": "CUST-002",
+      "title": "Blocked Sunday phone call",
+      "category": "custody",
+      "priority": 1,
+      "description": "Other parent blocked scheduled Sunday phone call with children.",
+      "sources": { "csv": true, "stickies": true, "timeline": true },
+      "timeline_events": [
+        {
+          "date": "2024-03-10",
+          "label": "Sunday call blocked",
+          "note": "Call time agreed; call not answered or blocked.",
+          "source": "timeline"
+        }
+      ],
+      "files": [
+        {
+          "path": "DivorceFiles/CL-xxx_2024-03-10_call-log.pdf",
+          "hash_sha256": "",
+          "exhibit_label": null
+        }
+      ],
+      "tags": ["phone_access", "interference", "pattern"]
+    }
+  ],
+  "stickies": [
+    {
+      "id": "STICKY-001",
+      "evidence_id": "CUST-002",
+      "date": "2024-03-10",
+      "note": "Wife blocked Sunday call.",
+      "theme": "custody",
+      "priority": 1
+    }
+  ],
+  "timeline": [
+    {
+      "id": "EVT-2024-03-10-CALL",
+      "date": "2024-03-10",
+      "label": "Sunday call blocked",
+      "category": "custody",
+      "priority": 1,
+      "details": "Scheduled Sunday phone contact did not occur.",
+      "evidence_ids": ["CUST-002"],
+      "source": "csv+sticky+timeline"
+    }
+  ],
+  "unreferenced_ids": [
+    {
+      "id": "CUST-801",
+      "reason": "Defined in CSV but not referenced in stickies or timeline."
+    }
+  ]
+}
\ No newline at end of file
diff --git a/case/evidence_index.schema.json b/case/evidence_index.schema.json
new file mode 100644
index 0000000..e4777be
--- /dev/null
+++ b/case/evidence_index.schema.json
@@ -0,0 +1,173 @@
+{
+  "$schema": "https://json-schema.org/draft/2020-12/schema",
+  "title": "ProSe Evidence Index",
+  "description": "Canonical evidence index schema for ProSe Case Manager.",
+  "type": "object",
+  "required": ["case_id", "evidence"],
+  "properties": {
+    "case_id": {
+      "type": "string",
+      "description": "Internal identifier for the case."
+    },
+    "generated_at": {
+      "type": "string",
+      "format": "date-time",
+      "description": "Timestamp when this index was generated."
+    },
+    "notes": {
+      "type": "string",
+      "description": "Optional human-readable notes about this snapshot."
+    },
+    "evidence": {
+      "type": "array",
+      "description": "Master list of evidence items for the case.",
+      "items": {
+        "type": "object",
+        "required": ["id", "title", "category"],
+        "properties": {
+          "id": {
+            "type": "string",
+            "description": "Evidence ID such as CUST-001 or SAFE-002."
+          },
+          "title": {
+            "type": "string",
+            "description": "Short title for this evidence item."
+          },
+          "category": {
+            "type": "string",
+            "description": "High-level category (e.g. custody, safety, procedural)."
+          },
+          "priority": {
+            "type": "number",
+            "description": "Priority score or tier. Lower = more important."
+          },
+          "description": {
+            "type": "string",
+            "description": "Longer factual description of what this evidence shows."
+          },
+          "sources": {
+            "type": "object",
+            "description": "Which input systems reference this ID.",
+            "properties": {
+              "csv": { "type": "boolean" },
+              "stickies": { "type": "boolean" },
+              "timeline": { "type": "boolean" }
+            },
+            "additionalProperties": false
+          },
+          "timeline_events": {
+            "type": "array",
+            "description": "Specific dated events associated with this evidence.",
+            "items": {
+              "type": "object",
+              "properties": {
+                "date": {
+                  "type": "string",
+                  "format": "date",
+                  "description": "YYYY-MM-DD"
+                },
+                "label": {
+                  "type": "string",
+                  "description": "Short label for the event."
+                },
+                "note": {
+                  "type": "string",
+                  "description": "Optional extra details about the event."
+                },
+                "source": {
+                  "type": "string",
+                  "description": "Where this event came from (timeline, sticky, etc.)."
+                }
+              },
+              "additionalProperties": false
+            }
+          },
+          "files": {
+            "type": "array",
+            "description": "Physical or digital files backing this evidence.",
+            "items": {
+              "type": "object",
+              "properties": {
+                "path": {
+                  "type": "string",
+                  "description": "Relative path to file (e.g. DivorceFiles/CL-001_xxx.pdf)."
+                },
+                "hash_sha256": {
+                  "type": "string",
+                  "description": "Optional SHA-256 hash for integrity checks."
+                },
+                "exhibit_label": {
+                  "type": ["string", "null"],
+                  "description": "Court exhibit label once assigned."
+                }
+              },
+              "additionalProperties": false
+            }
+          },
+          "tags": {
+            "type": "array",
+            "description": "Searchable tags summarizing themes or issues.",
+            "items": { "type": "string" }
+          }
+        },
+        "additionalProperties": false
+      }
+    },
+    "stickies": {
+      "type": "array",
+      "description": "Sticky-note style fact snippets linked to evidence IDs.",
+      "items": {
+        "type": "object",
+        "properties": {
+          "id": { "type": "string" },
+          "evidence_id": { "type": "string" },
+          "date": {
+            "type": "string",
+            "format": "date"
+          },
+          "note": { "type": "string" },
+          "theme": { "type": "string" },
+          "priority": { "type": "number" }
+        },
+        "additionalProperties": false
+      }
+    },
+    "timeline": {
+      "type": "array",
+      "description": "Chronological events referencing evidence IDs.",
+      "items": {
+        "type": "object",
+        "properties": {
+          "id": { "type": "string" },
+          "date": {
+            "type": "string",
+            "format": "date"
+          },
+          "label": { "type": "string" },
+          "category": { "type": "string" },
+          "priority": { "type": "number" },
+          "details": { "type": "string" },
+          "evidence_ids": {
+            "type": "array",
+            "items": { "type": "string" }
+          },
+          "source": { "type": "string" }
+        },
+        "additionalProperties": false
+      }
+    },
+    "unreferenced_ids": {
+      "type": "array",
+      "description": "Evidence IDs defined but not referenced anywhere yet.",
+      "items": {
+        "type": "object",
+        "properties": {
+          "id": { "type": "string" },
+          "reason": { "type": "string" }
+        },
+        "additionalProperties": false
+      }
+    }
+  },
+  "additionalProperties": false
+}
\ No newline at end of file
diff --git a/docs/EVIDENCE_INDEX.md b/docs/EVIDENCE_INDEX.md
new file mode 100644
index 0000000..6f3f964
--- /dev/null
+++ b/docs/EVIDENCE_INDEX.md
@@ -0,0 +1,95 @@
+# ProSe Evidence Index
+
+The Evidence Index is the **single source of truth** for a case’s structured evidence in ProSe.
+
+It combines:
+
+- the master evidence list (`Custody_Mod_Evidence.csv`)
+- sticky-note style facts (`sticky_index.json`)
+- timeline events (`timeline.csv`)
+
+into one canonical JSON document that the Case Manager can safely consume.
+
+---
+
+## Goals
+
+- Ensure every `Evidence_ID` is:
+  - declared once,
+  - referenced consistently,
+  - and never silently “drifts.”
+- Provide a stable input for:
+  - timeline generation,
+  - affidavit drafting,
+  - motion templates,
+  - and audit reports.
+
+---
+
+## Files
+
+- `case/evidence_index.schema.json`
+  JSON Schema definition for the index.
+
+- `case/evidence_index.sample.json`
+  Example instance of the index, populated with sample data.
+
+- `engine/agents/evidence_validator.py`
+  Module that cross-checks the CSV/JSON/timeline source files.
+
+---
+
+## Source Inputs
+
+The validator expects a case directory (e.g. `my_real_case/`) containing:
+
+- `Custody_Mod_Evidence.csv`
+  - Must have a column: `Evidence_ID`
+- `sticky_index.json`
+  - List of objects, each with `evidence_ids: [ ... ]`
+- `timeline.csv`
+  - Column `Evidence_IDs` with `;`-separated IDs
+
+These three are treated as the “donor truth” that gets normalized into the Evidence Index.
+
+---
+
+## Validation Logic
+
+For a given case directory:
+
+1. Load all `Evidence_ID` values from the CSV.
+2. Load all `evidence_ids` from stickies.
+3. Load all `Evidence_IDs` from the timeline.
+4. Compute:
+   - IDs used in stickies but not in CSV (`unknown_in_stickies`)
+   - IDs used in timeline but not in CSV (`unknown_in_timeline`)
+   - IDs in CSV but not referenced anywhere (`unused_evidence`)
+5. Return a structured result and human-readable report.
+
+If there are no unknown IDs in stickies or timeline, status is `OK`.
+Otherwise, status is `WARN`.
+
+---
+
+## Case Manager Integration
+
+The Case Manager can:
+
+- Call `validate_case(base_path)` from `engine.agents.evidence_validator`.
+- Inspect the returned dict to:
+  - block or warn before generating court-facing documents,
+  - highlight missing or inconsistent IDs,
+  - propose follow-up tasks (“map CUST-801 into the timeline”).
+
+Later, the same data can be used to **export a full `evidence_index.json`** instance that conforms to `evidence_index.schema.json`.
+
+---
+
+## Next Steps
+
+- Implement an exporter that writes a valid `evidence_index.json` from the three source files.
+- Add tests where:
+  - CSV, stickies, and timeline all agree (status `OK`).
+  - Known mismatches are present (status `WARN`) and are correctly reported.
+- Wire this into the Case Manager’s “pre-flight check” before drafting motions or affidavits.
diff --git a/engine/__init__.py b/engine/__init__.py
new file mode 100644
index 0000000..e69de29
diff --git a/engine/agents/__init__.py b/engine/agents/__init__.py
new file mode 100644
index 0000000..e69de29
diff --git a/engine/agents/evidence_validator.py b/engine/agents/evidence_validator.py
new file mode 100644
index 0000000..5c698fe
--- /dev/null
+++ b/engine/agents/evidence_validator.py
@@ -0,0 +1,166 @@
+"""
+Evidence validator for ProSe.
+
+Cross-checks:
+- Custody_Mod_Evidence.csv (master evidence list)
+- sticky_index.json        (sticky notes referencing evidence_ids)
+- timeline.csv             (timeline events with Evidence_IDs field)
+
+and reports:
+- unknown IDs in stickies/timeline
+- unused evidence IDs in the CSV
+"""
+
+from __future__ import annotations
+
+import csv
+import json
+from dataclasses import dataclass, asdict
+from pathlib import Path
+from typing import Dict, List, Set
+
+
+CUSTODY_CSV_NAME = "Custody_Mod_Evidence.csv"
+STICKY_JSON_NAME = "sticky_index.json"
+TIMELINE_CSV_NAME = "timeline.csv"
+
+
+@dataclass
+class EvidenceValidationResult:
+    evidence_count: int
+    sticky_count: int
+    timeline_count: int
+    unknown_in_stickies: List[str]
+    unknown_in_timeline: List[str]
+    unused_evidence: List[str]
+    status: str  # "OK" or "WARN"
+
+
+def _load_evidence_ids(base: Path) -> Set[str]:
+    path = base / CUSTODY_CSV_NAME
+    ids: Set[str] = set()
+
+    if not path.exists():
+        print(f"[evidence_validator] Warning: {path} not found")
+        return ids
+
+    with path.open(newline="", encoding="utf-8") as f:
+        reader = csv.DictReader(f)
+        for row in reader:
+            eid = (row.get("Evidence_ID") or "").strip()
+            if eid:
+                ids.add(eid)
+    return ids
+
+
+def _load_sticky_ids(base: Path) -> Set[str]:
+    path = base / STICKY_JSON_NAME
+    ids: Set[str] = set()
+
+    if not path.exists():
+        return ids
+
+    data = json.loads(path.read_text(encoding="utf-8"))
+    # expect a list of objects with "evidence_ids": [...]
+    for sticky in data:
+        for eid in sticky.get("evidence_ids", []):
+            eid_clean = (eid or "").strip()
+            if eid_clean:
+                ids.add(eid_clean)
+    return ids
+
+
+def _load_timeline_ids(base: Path) -> Set[str]:
+    path = base / TIMELINE_CSV_NAME
+    ids: Set[str] = set()
+
+    if not path.exists():
+        return ids
+
+    with path.open(newline="", encoding="utf-8") as f:
+        reader = csv.DictReader(f)
+        for row in reader:
+            field = row.get("Evidence_IDs") or ""
+            for eid in [x.strip() for x in field.split(";") if x.strip()]:
+                ids.add(eid)
+    return ids
+
+
+def validate_case(base: Path) -> Dict[str, object]:
+    """
+    Validate evidence links for a given case directory.
+
+    :param base: Path to directory containing Custody_Mod_Evidence.csv,
+                 sticky_index.json, timeline.csv
+    :return: dict suitable for JSON or further processing.
+    """
+    base = Path(base)
+
+    evidence_ids = _load_evidence_ids(base)
+    sticky_ids = _load_sticky_ids(base)
+    timeline_ids = _load_timeline_ids(base)
+
+    unknown_in_stickies = sticky_ids - evidence_ids
+    unknown_in_timeline = timeline_ids - evidence_ids
+    unused_evidence = evidence_ids - sticky_ids - timeline_ids
+
+    status = "OK"
+    if unknown_in_stickies or unknown_in_timeline:
+        status = "WARN"
+
+    result = EvidenceValidationResult(
+        evidence_count=len(evidence_ids),
+        sticky_count=len(sticky_ids),
+        timeline_count=len(timeline_ids),
+        unknown_in_stickies=sorted(unknown_in_stickies),
+        unknown_in_timeline=sorted(unknown_in_timeline),
+        unused_evidence=sorted(unused_evidence),
+        status=status,
+    )
+    return asdict(result)
+
+
+def print_report(result: Dict[str, object]) -> None:
+    """
+    Pretty-print validation results to the console.
+    """
+    print(f"Evidence IDs in CSV: {result['evidence_count']}")
+    print(f"Referenced in stickies: {result['sticky_count']}")
+    print(f"Referenced in timeline: {result['timeline_count']}")
+
+    unknown_in_stickies = result["unknown_in_stickies"]
+    unknown_in_timeline = result["unknown_in_timeline"]
+    unused_evidence = result["unused_evidence"]
+
+    if unknown_in_stickies:
+        print("\n⚠ Unknown Evidence_IDs in stickies (not in CSV):")
+        for eid in unknown_in_stickies:
+            print(f"  - {eid}")
+
+    if unknown_in_timeline:
+        print("\n⚠ Unknown Evidence_IDs in timeline (not in CSV):")
+        for eid in unknown_in_timeline:
+            print(f"  - {eid}")
+
+    if unused_evidence:
+        print("\nℹ Evidence_IDs in CSV not referenced yet (fine, but FYI):")
+        for eid in unused_evidence:
+            print(f"  - {eid}")
+
+    if not (unknown_in_stickies or unknown_in_timeline):
+        print("\n✅ Links look consistent. Nice work.")
+    else:
+        print("\n⚠ Validation completed with warnings. Review above items.")
+
+
+if __name__ == "__main__":
+    # CLI usage: python -m engine.agents.evidence_validator my_real_case
+    import sys
+
+    if len(sys.argv) > 1:
+        base_dir = Path(sys.argv[1])
+    else:
+        base_dir = Path("my_real_case")
+
+    result_dict = validate_case(base_dir)
+    print_report(result_dict)
diff --git a/tests/core/test_evidence_validator.py b/tests/core/test_evidence_validator.py
new file mode 100644
index 0000000..fd45837
--- /dev/null
+++ b/tests/core/test_evidence_validator.py
@@ -0,0 +1,20 @@
+from pathlib import Path
+
+from engine.agents.evidence_validator import validate_case
+
+
+def test_empty_case_directory(tmp_path: Path) -> None:
+    """
+    With no CSV/JSON/Timeline files present,
+    the validator should not crash and should
+    report zero counts and OK status.
+    """
+    result = validate_case(tmp_path)
+
+    assert result["evidence_count"] == 0
+    assert result["sticky_count"] == 0
+    assert result["timeline_count"] == 0
+    assert result["unknown_in_stickies"] == []
+    assert result["unknown_in_timeline"] == []
+    assert result["unused_evidence"] == []
+    assert result["status"] == "OK"