diff --git a/Makefile b/Makefile
new file mode 100644
index 0000000000..641828b301
--- /dev/null
+++ b/Makefile
@@ -0,0 +1,12 @@
+# Republic OS — repeatable pipeline commands.
+# The build is deterministic: `make build` twice yields a byte-identical tree.
+
+.PHONY: build validate check
+
+build:        ## Regenerate the entity tree from raw exports in data/
+	python3 scripts/build.py
+
+validate:     ## Check every entity file: OKF, schema, and link integrity
+	python3 scripts/validate.py
+
+check: build validate  ## Build, then validate — the full gate
diff --git a/schemas/body.schema.json b/schemas/body.schema.json
new file mode 100644
index 0000000000..243d601586
--- /dev/null
+++ b/schemas/body.schema.json
@@ -0,0 +1,20 @@
+{
+  "$schema": "http://json-schema.org/draft-07/schema#",
+  "title": "Body",
+  "description": "A government body or institution — a chamber, committee, or subcommittee. The 'repos' of the government-as-code model.",
+  "type": "object",
+  "required": ["type", "title", "classification", "chamber", "code", "confidence", "tags", "timestamp"],
+  "properties": {
+    "type": { "type": "string", "enum": ["Body"] },
+    "title": { "type": "string" },
+    "classification": { "type": "string", "enum": ["house", "senate", "executive", "committee", "subcommittee"] },
+    "chamber": { "type": "string", "enum": ["house", "senate", "joint", "executive"] },
+    "code": { "type": "string" },
+    "parent": { "type": "string" },
+    "leadership": { "type": "array" },
+    "sources": { "type": "array" },
+    "confidence": { "type": "string", "enum": ["official", "reported", "inferred", "unverified"] },
+    "tags": { "type": "array" },
+    "timestamp": { "type": "string" }
+  }
+}
diff --git a/schemas/candidate.schema.json b/schemas/candidate.schema.json
new file mode 100644
index 0000000000..fbe83b8e8e
--- /dev/null
+++ b/schemas/candidate.schema.json
@@ -0,0 +1,24 @@
+{
+  "$schema": "http://json-schema.org/draft-07/schema#",
+  "title": "Candidate",
+  "description": "A candidate for public office — someone running, distinct from a current officeholder. Sourced from FEC filings.",
+  "type": "object",
+  "required": ["type", "title", "office", "state", "confidence", "tags", "timestamp", "ids"],
+  "properties": {
+    "type": { "type": "string", "enum": ["Candidate"] },
+    "title": { "type": "string" },
+    "office": { "type": "string", "enum": ["U.S. House", "U.S. Senate"] },
+    "state": { "type": "string" },
+    "district": { "type": "string" },
+    "party": { "type": "string" },
+    "stance": { "type": "string", "enum": ["incumbent", "challenger", "open-seat"] },
+    "election_year": { "type": "integer" },
+    "status": { "type": "string" },
+    "committee": { "type": "object" },
+    "ids": { "type": "object" },
+    "sources": { "type": "array" },
+    "confidence": { "type": "string", "enum": ["official", "reported", "inferred", "unverified"] },
+    "tags": { "type": "array" },
+    "timestamp": { "type": "string" }
+  }
+}
diff --git a/schemas/jurisdiction.schema.json b/schemas/jurisdiction.schema.json
new file mode 100644
index 0000000000..3fb078a059
--- /dev/null
+++ b/schemas/jurisdiction.schema.json
@@ -0,0 +1,20 @@
+{
+  "$schema": "http://json-schema.org/draft-07/schema#",
+  "title": "Jurisdiction",
+  "description": "A place with a government — a county or congressional district — carrying demographic context. A node in the fractal jurisdiction tree.",
+  "type": "object",
+  "required": ["type", "title", "classification", "confidence", "tags", "timestamp"],
+  "properties": {
+    "type": { "type": "string", "enum": ["Jurisdiction"] },
+    "title": { "type": "string" },
+    "classification": { "type": "string", "enum": ["county", "congressional-district"] },
+    "fips": { "type": "string" },
+    "state": { "type": "string" },
+    "district": { "type": "string" },
+    "demographics": { "type": "object" },
+    "sources": { "type": "array" },
+    "confidence": { "type": "string", "enum": ["official", "reported", "inferred", "unverified"] },
+    "tags": { "type": "array" },
+    "timestamp": { "type": "string" }
+  }
+}
diff --git a/schemas/person.schema.json b/schemas/person.schema.json
new file mode 100644
index 0000000000..696a31f472
--- /dev/null
+++ b/schemas/person.schema.json
@@ -0,0 +1,23 @@
+{
+  "$schema": "http://json-schema.org/draft-07/schema#",
+  "title": "Person",
+  "description": "An officeholder — a person holding a public office at any level. OKF-conformant: only `type` is strictly required by OKF; this profile adds the fields Republic OS relies on. Unknown keys are tolerated.",
+  "type": "object",
+  "required": ["type", "title", "level", "confidence", "tags", "timestamp", "ids"],
+  "properties": {
+    "type": { "type": "string", "enum": ["Person"] },
+    "title": { "type": "string" },
+    "role": { "type": "string" },
+    "party": { "type": "string" },
+    "level": { "type": "string", "enum": ["federal", "state", "county", "municipal"] },
+    "branch": { "type": "string", "enum": ["executive", "legislative", "judicial"] },
+    "state": { "type": "string" },
+    "leadership": { "type": "array" },
+    "committees": { "type": "array" },
+    "ids": { "type": "object" },
+    "sources": { "type": "array" },
+    "confidence": { "type": "string", "enum": ["official", "reported", "inferred", "unverified"] },
+    "tags": { "type": "array" },
+    "timestamp": { "type": "string" }
+  }
+}
diff --git a/scripts/validate.py b/scripts/validate.py
new file mode 100644
index 0000000000..d688b2985e
--- /dev/null
+++ b/scripts/validate.py
@@ -0,0 +1,205 @@
+#!/usr/bin/env python3
+"""Validate every entity file in the mirror.
+
+Three checks, no third-party dependencies:
+  1. OKF conformance — every non-reserved .md has YAML frontmatter with a
+     non-empty `type`.
+  2. Schema conformance — each entity validates against the JSON Schema for
+     its `type` in /schemas (required fields, property types, enums). The
+     schemas are a stricter profile over OKF's permissive base; unknown keys
+     are tolerated by design.
+  3. Link integrity — internal /-rooted markdown links resolve to a file that
+     exists in the tree.
+
+Exit code is non-zero if any check fails, so this doubles as a CI gate.
+
+Usage:  python scripts/validate.py
+"""
+
+import json
+import re
+import sys
+from pathlib import Path
+
+REPO = Path(__file__).resolve().parent.parent
+TREE = REPO / "data" / "jurisdictions"
+SCHEMAS = REPO / "schemas"
+MAX_REPORT = 25
+
+TYPE_PY = {
+    "string": str, "number": (int, float), "integer": int,
+    "boolean": bool, "array": list, "object": dict,
+}
+
+
+# --------------------------------------------------------------------------- #
+# frontmatter parser (tuned to the deterministic output of build.py)
+# --------------------------------------------------------------------------- #
+
+def extract_frontmatter(text):
+    lines = text.split("\n")
+    if not lines or lines[0].strip() != "---":
+        return None
+    out = []
+    for line in lines[1:]:
+        if line.strip() == "---":
+            return out
+        out.append(line)
+    return None
+
+
+def parse_scalar(s):
+    s = s.strip()
+    if s == "":
+        return None
+    try:
+        return json.loads(s)          # "quoted", 123, 1.5, true, false, null
+    except (ValueError, json.JSONDecodeError):
+        return s                       # bare word, e.g. official
+
+
+def parse_value(s):
+    s = s.strip()
+    if s.startswith("[") and s.endswith("]"):
+        inner = s[1:-1].strip()
+        return [parse_scalar(x) for x in inner.split(",")] if inner else []
+    return parse_scalar(s)
+
+
+def parse_frontmatter(text):
+    """Top-level parse. Block values become [] (array) or {} (object)
+    sentinels — sufficient for schema shape/enum/required checks."""
+    fm = extract_frontmatter(text)
+    if fm is None:
+        return None
+    d = {}
+    i = 0
+    while i < len(fm):
+        line = fm[i]
+        if not line.strip() or line.startswith((" ", "\t")):
+            i += 1
+            continue
+        m = re.match(r"([A-Za-z0-9_]+):(.*)$", line)
+        if not m:
+            i += 1
+            continue
+        key, rest = m.group(1), m.group(2).strip()
+        if rest:
+            d[key] = parse_value(rest)
+        else:
+            j = i + 1
+            while j < len(fm) and not fm[j].strip():
+                j += 1
+            d[key] = [] if (j < len(fm) and fm[j].lstrip().startswith("- ")) else {}
+        i += 1
+    return d
+
+
+# --------------------------------------------------------------------------- #
+# minimal JSON-Schema validation (required / type / enum)
+# --------------------------------------------------------------------------- #
+
+def validate_obj(obj, schema):
+    errs = []
+    for req in schema.get("required", []):
+        if req not in obj or obj[req] is None:
+            errs.append(f"missing required '{req}'")
+    for key, spec in schema.get("properties", {}).items():
+        if key not in obj or obj[key] is None:
+            continue
+        val = obj[key]
+        t = spec.get("type")
+        if t and not isinstance(val, TYPE_PY[t]):
+            # bool is a subclass of int; reject bool where integer/number expected
+            if not (t in ("integer", "number") and isinstance(val, bool) is False):
+                errs.append(f"'{key}' expected {t}, got {type(val).__name__}")
+        if "enum" in spec and val not in spec["enum"]:
+            errs.append(f"'{key}'={val!r} not in {spec['enum']}")
+    return errs
+
+
+# --------------------------------------------------------------------------- #
+# main
+# --------------------------------------------------------------------------- #
+
+def load_schemas():
+    schemas = {}
+    for p in sorted(SCHEMAS.glob("*.schema.json")):
+        s = json.loads(p.read_text())
+        t = s["properties"]["type"]["enum"][0]
+        schemas[t] = s
+    return schemas
+
+
+LINK_RE = re.compile(r"\]\((/[^)\s]+\.md)\)")
+
+
+def main():
+    schemas = load_schemas()
+    if not schemas:
+        print("no schemas found in /schemas", file=sys.stderr)
+        return 1
+
+    files = sorted(TREE.rglob("*.md"))
+    counts = {t: 0 for t in schemas}
+    counts["(untyped)"] = 0
+    okf_fail, schema_fail, link_fail = [], [], []
+    existing = {("/" + str(p.relative_to(TREE))) for p in files}
+
+    for p in files:
+        rel = p.relative_to(REPO)
+        text = p.read_text(encoding="utf-8")
+        fm = parse_frontmatter(text)
+
+        # 1. OKF conformance
+        if fm is None or not fm.get("type"):
+            okf_fail.append(f"{rel}: no frontmatter / empty type")
+            counts["(untyped)"] += 1
+            continue
+
+        etype = fm["type"]
+        # 2. schema conformance
+        if etype in schemas:
+            counts[etype] += 1
+            for e in validate_obj(fm, schemas[etype]):
+                schema_fail.append(f"{rel}: {e}")
+        else:
+            schema_fail.append(f"{rel}: unknown type {etype!r} (no schema)")
+
+        # 3. link integrity
+        for m in LINK_RE.finditer(text):
+            if m.group(1) not in existing:
+                link_fail.append(f"{rel}: dead link {m.group(1)}")
+
+    # ---- report ----
+    total = len(files)
+    print(f"validated {total} entity files")
+    for t in sorted(counts):
+        if counts[t]:
+            print(f"  {t}: {counts[t]}")
+
+    def section(name, items):
+        if not items:
+            print(f"\n{name}: PASS (0)")
+            return 0
+        print(f"\n{name}: FAIL ({len(items)})")
+        for line in items[:MAX_REPORT]:
+            print(f"  - {line}")
+        if len(items) > MAX_REPORT:
+            print(f"  … and {len(items) - MAX_REPORT} more")
+        return len(items)
+
+    fails = (section("OKF conformance", okf_fail)
+             + section("Schema conformance", schema_fail)
+             + section("Link integrity", link_fail))
+
+    print()
+    if fails:
+        print(f"VALIDATION FAILED — {fails} issue(s)")
+        return 1
+    print("VALIDATION PASSED — all files conform")
+    return 0
+
+
+if __name__ == "__main__":
+    sys.exit(main())