fix(protocol): harden gateway contract validation

tools/check_scope.py

@@ -30,6 +30,9 @@ SECRET_PATTERNS = (
     re.compile(rb"\bgh[pousr]_[A-Za-z0-9]{20,}\b"),
     re.compile(rb"\bsk-[A-Za-z0-9]{20,}\b"),
 )
+ALLOWED_SECRET_PROPERTIES = {
+    ("ProviderSessionWork", "client_private_key_pem"),
+}
 
 
 def fail(message: str) -> None:
@@ -53,19 +56,36 @@ def check_generated_provenance() -> None:
 def check_manifest_schema() -> None:
     schema = json.loads((ROOT / "schemas/control-v1.schema.json").read_text(encoding="utf-8"))
     definitions = schema.get("$defs", {})
-    for name in ("ConnectionManifest", "ManifestGateway", "ManifestTunnel", "ManifestProfile", "ManifestBounds", "GrantReference"):
-        properties = definitions.get(name, {}).get("properties", {})
-        forbidden = sorted(FORBIDDEN_WIRE_FIELDS.intersection(properties))
-        if forbidden:
-            fail(f"{name} exposes forbidden wire fields: {forbidden}")
+    for name, definition in definitions.items():
+        for field in definition.get("properties", {}):
+            if any(forbidden in field.lower() for forbidden in FORBIDDEN_WIRE_FIELDS):
+                if (name, field) not in ALLOWED_SECRET_PROPERTIES:
+                    fail(f"{name} exposes forbidden wire field {field}")
+
+
+def check_proto_boundaries(path: pathlib.Path) -> None:
+    message = ""
+    depth = 0
+    for line in path.read_text(encoding="utf-8").splitlines():
+        match = re.match(r"\s*message\s+([A-Za-z0-9_]+)\s*\{", line)
+        if match and depth == 0:
+            message = match.group(1)
+        if any(field in line.lower() for field in FORBIDDEN_WIRE_FIELDS):
+            allowed = (
+                message == "ProviderSessionWork"
+                and re.fullmatch(r"\s*string\s+client_private_key_pem\s*=\s*[0-9]+;\s*", line)
+            )
+            if not allowed:
+                fail(f"{path.relative_to(ROOT)} exposes forbidden wire field in {message or 'file scope'}")
+        depth += line.count("{") - line.count("}")
+        if depth == 0:
+            message = ""
 
 
 def check_text_boundaries() -> None:
     paths = [
         ROOT / "openapi/control-v1.yaml",
-        ROOT / "schemas/control-v1.schema.json",
         ROOT / "proto/versevdi/control/v1/control.proto",
-        ROOT / "proto/versevdi/tunnel/v1/tunnel.proto",
         ROOT / "frames/datagram-v1.md",
         ROOT / "frames/registry.json",
         ROOT / "registries/features.json",
@@ -77,14 +97,7 @@ def check_text_boundaries() -> None:
             if field in text:
                 fail(f"{path.relative_to(ROOT)} contains forbidden wire field {field}")
 
-    generated_paths = list((ROOT / "gen").rglob("*"))
-    for path in generated_paths:
-        if not path.is_file() or path.name == "manifest.json" or path.suffix in {".pb", ".binpb"}:
-            continue
-        text = path.read_text(encoding="utf-8").lower()
-        for field in FORBIDDEN_WIRE_FIELDS:
-            if field in text:
-                fail(f"generated output {path.relative_to(ROOT)} contains forbidden wire field {field}")
+    check_proto_boundaries(ROOT / "proto/versevdi/tunnel/v1/tunnel.proto")
 
 
 def check_secret_canaries() -> None:

tools/generate.py

@@ -125,6 +125,8 @@ def go_validation(definition: dict[str, Any]) -> list[str]:
                 lines.append(f"\tif len(v.{field}) < {prop['minLength']} && v.{field} != \"\" {{ violations = append(violations, FieldViolation{{Field: \"{prop_name}\", Code: \"min_length\"}}) }}")
             if "maxLength" in prop:
                 lines.append(f"\tif len(v.{field}) > {prop['maxLength']} {{ violations = append(violations, FieldViolation{{Field: \"{prop_name}\", Code: \"max_length\"}}) }}")
+            if "x-max-bytes" in prop:
+                lines.append(f"\tif len(v.{field}) > {prop['x-max-bytes']} {{ violations = append(violations, FieldViolation{{Field: \"{prop_name}\", Code: \"max_bytes\"}}) }}")
             if "const" in prop:
                 lines.append(f"\tif v.{field} != \"{prop['const']}\" && v.{field} != \"\" {{ violations = append(violations, FieldViolation{{Field: \"{prop_name}\", Code: \"invalid_value\"}}) }}")
             if "enum" in prop:
@@ -135,6 +137,8 @@ def go_validation(definition: dict[str, Any]) -> list[str]:
                     '\tif v.%s != "" { if parsed, err := time.Parse(time.RFC3339Nano, v.%s); err != nil || parsed.UTC().Format(time.RFC3339Nano) != v.%s { violations = append(violations, FieldViolation{Field: "%s", Code: "invalid_time"}) } }'
                     % (field, field, field, prop_name)
                 )
+            if prop.get("format") == "base64url":
+                lines.append(f"\tif v.{field} != \"\" {{ if _, err := base64.RawURLEncoding.Strict().DecodeString(v.{field}); err != nil {{ violations = append(violations, FieldViolation{{Field: \"{prop_name}\", Code: \"invalid_format\"}}) }} }}")
         if prop.get("type") == "integer":
             if "minimum" in prop:
                 lines.append(f"\tif v.{field} != 0 && v.{field} < {prop['minimum']} {{ violations = append(violations, FieldViolation{{Field: \"{prop_name}\", Code: \"minimum\"}}) }}")
@@ -167,6 +171,7 @@ def generate_go(defs: dict[str, dict[str, Any]], schema_hash: str, version: str,
         "",
         "import (",
         "\"bytes\"",
+        "\"encoding/base64\"",
         "\"encoding/json\"",
         "\"errors\"",
         "\"fmt\"",
@@ -223,7 +228,7 @@ def generate_go(defs: dict[str, dict[str, Any]], schema_hash: str, version: str,
                 % (prop_name, prop_name)
             )
         for prop_name, prop in defs[name].get("properties", {}).items():
-            if "x-max-bytes" in prop:
+            if "x-max-bytes" in prop and prop.get("type") != "string":
                 out.append(
                     '\tif raw, ok := fields["%s"]; ok && len(raw) > %d { return value, ValidationError{Violations: []FieldViolation{{Field: "%s", Code: "max_bytes"}}} }'
                     % (prop_name, prop["x-max-bytes"], prop_name)
@@ -322,11 +327,15 @@ def rust_validation(definition: dict[str, Any]) -> list[str]:
                 lines.append(f"        {prefix}if !{value}.is_empty() && {value}.len() < {prop['minLength']} {{ return Err(ValidationError::new(\"{prop_name}\", \"min_length\")); }}")
             if "maxLength" in prop:
                 lines.append(f"        {prefix}if {value}.len() > {prop['maxLength']} {{ return Err(ValidationError::new(\"{prop_name}\", \"max_length\")); }}")
+            if "x-max-bytes" in prop:
+                lines.append(f"        {prefix}if {value}.as_bytes().len() > {prop['x-max-bytes']} {{ return Err(ValidationError::new(\"{prop_name}\", \"max_bytes\")); }}")
             if "const" in prop:
                 lines.append(f"        {prefix}if {value} != \"{prop['const']}\" {{ return Err(ValidationError::new(\"{prop_name}\", \"invalid_value\")); }}")
             if "enum" in prop:
                 allowed = " && ".join(f'{value} != \"{item}\"' for item in prop["enum"])
                 lines.append(f"        {prefix}if {allowed} {{ return Err(ValidationError::new(\"{prop_name}\", \"invalid_value\")); }}")
+            if prop.get("format") == "base64url":
+                lines.append(f"        {prefix}if !valid_base64_url({value}.as_str()) {{ return Err(ValidationError::new(\"{prop_name}\", \"invalid_format\")); }}")
         if prop.get("type") == "integer":
             if "minimum" in prop:
                 lines.append(f"        {prefix}if {value} < {prop['minimum']} {{ return Err(ValidationError::new(\"{prop_name}\", \"minimum\")); }}")
@@ -369,6 +378,27 @@ def generate_rust(defs: dict[str, dict[str, Any]], schema_hash: str, compatibili
         "pub struct ValidationError { pub field: &'static str, pub code: &'static str }",
         "impl ValidationError { pub const fn new(field: &'static str, code: &'static str) -> Self { Self { field, code } } }",
         "",
+        "fn base64url_value(value: u8) -> Option<u8> {",
+        "    match value {",
+        "        b'A'..=b'Z' => Some(value - b'A'),",
+        "        b'a'..=b'z' => Some(value - b'a' + 26),",
+        "        b'0'..=b'9' => Some(value - b'0' + 52),",
+        "        b'-' => Some(62),",
+        "        b'_' => Some(63),",
+        "        _ => None,",
+        "    }",
+        "}",
+        "fn valid_base64_url(value: &str) -> bool {",
+        "    let bytes = value.as_bytes();",
+        "    if bytes.is_empty() || bytes.iter().any(|byte| base64url_value(*byte).is_none()) { return false; }",
+        "    match bytes.len() % 4 {",
+        "        0 => true,",
+        "        2 => base64url_value(*bytes.last().unwrap()).unwrap() & 0x0f == 0,",
+        "        3 => base64url_value(*bytes.last().unwrap()).unwrap() & 0x03 == 0,",
+        "        _ => false,",
+        "    }",
+        "}",
+        "",
     ]
     for name in sorted(defs):
         definition = defs[name]
@@ -451,6 +481,8 @@ def swift_validation(definition: dict[str, Any]) -> list[str]:
                 lines.append(f"        {prefix}if !{value}.isEmpty && {value}.utf8.count < {prop['minLength']} {{ throw ContractValidationError(field: \"{prop_name}\", code: \"min_length\") }}")
             if "maxLength" in prop:
                 lines.append(f"        {prefix}if {value}.utf8.count > {prop['maxLength']} {{ throw ContractValidationError(field: \"{prop_name}\", code: \"max_length\") }}")
+            if "x-max-bytes" in prop:
+                lines.append(f"        {prefix}if {value}.utf8.count > {prop['x-max-bytes']} {{ throw ContractValidationError(field: \"{prop_name}\", code: \"max_bytes\") }}")
             if "const" in prop:
                 lines.append(f"        {prefix}if {value} != \"{prop['const']}\" {{ throw ContractValidationError(field: \"{prop_name}\", code: \"invalid_value\") }}")
             if "enum" in prop:
@@ -458,6 +490,8 @@ def swift_validation(definition: dict[str, Any]) -> list[str]:
                 lines.append(f"        {prefix}if ![{allowed}].contains({value}) {{ throw ContractValidationError(field: \"{prop_name}\", code: \"invalid_value\") }}")
             if prop.get("format") == "date-time":
                 lines.append(f"        {prefix}if ISO8601DateFormatter().date(from: {value}) == nil {{ throw ContractValidationError(field: \"{prop_name}\", code: \"invalid_time\") }}")
+            if prop.get("format") == "base64url":
+                lines.append(f"        {prefix}if !validBase64URL({value}) {{ throw ContractValidationError(field: \"{prop_name}\", code: \"invalid_format\") }}")
         if prop.get("type") == "integer":
             if "minimum" in prop:
                 lines.append(f"        {prefix}if {value} < {prop['minimum']} {{ throw ContractValidationError(field: \"{prop_name}\", code: \"minimum\") }}")
@@ -497,6 +531,15 @@ def generate_swift(defs: dict[str, dict[str, Any]], schema_hash: str, compatibil
         f'public let nMinus2WireVersion = "{compatibility["n_minus_2"]}"',
         "public struct ContractValidationError: Error, Equatable { public let field: String; public let code: String }",
         "private struct AnyCodingKey: CodingKey { let stringValue: String; let intValue: Int?; init?(stringValue: String) { self.stringValue = stringValue; self.intValue = nil }; init?(intValue: Int) { self.stringValue = String(intValue); self.intValue = intValue } }",
+        "private func validBase64URL(_ value: String) -> Bool {",
+        "    guard !value.isEmpty, value.utf8.allSatisfy({ byte in",
+        "        (byte >= 65 && byte <= 90) || (byte >= 97 && byte <= 122) || (byte >= 48 && byte <= 57) || byte == 45 || byte == 95",
+        "    }) else { return false }",
+        "    let padding = String(repeating: \"=\", count: (4 - value.utf8.count % 4) % 4)",
+        "    let standard = value.replacingOccurrences(of: \"-\", with: \"+\").replacingOccurrences(of: \"_\", with: \"/\") + padding",
+        "    guard let decoded = Data(base64Encoded: standard) else { return false }",
+        "    return decoded.base64EncodedString().replacingOccurrences(of: \"+\", with: \"-\").replacingOccurrences(of: \"/\", with: \"_\").replacingOccurrences(of: \"=\", with: \"\") == value",
+        "}",
         "",
     ]
     for name in sorted(defs):

tools/native_conformance.rs

@@ -56,12 +56,18 @@ fn evaluate(kind: &str, input: &str) -> &'static str {
         "gateway_input" => classify_gateway_input(values.get("hex").map(String::as_str).unwrap_or_default()),
         "gateway_feedback" => classify_gateway_feedback(values.get("hex").map(String::as_str).unwrap_or_default()),
         "gateway_clipboard" if values.contains_key("file") => "invalid:forbidden",
-        "gateway_clipboard"
-            if matches!(values.get("direction").map(String::as_str), Some("client_to_provider") | Some("provider_to_client"))
-                && values.get("encoding").map(String::as_str) == Some("utf-8")
-                && values.get("loop_token").map_or(false, |value| (16..=128).contains(&value.len()))
-                && values.get("text").map_or(false, |value| value.len() <= 65536) => "valid",
-        "gateway_clipboard" => "invalid:clipboard",
+        "gateway_clipboard" => match (
+            values.get("direction"),
+            values.get("text"),
+            values.get("encoding"),
+            values.get("loop_token"),
+        ) {
+            (Some(direction), Some(text), Some(encoding), Some(token))
+                if GatewayClipboardText::new(
+                    direction.clone(), text.clone(), encoding.clone(), token.clone(),
+                ).is_ok() => "valid",
+            _ => "invalid:clipboard",
+        },
         "gateway_clipboard_audit" if values.contains_key("text") => "invalid:forbidden",
         "gateway_clipboard_audit"
             if matches!(values.get("direction").map(String::as_str), Some("client_to_provider") | Some("provider_to_client"))

tools/native_conformance.swift

@@ -34,7 +34,11 @@ func evaluate(_ kind: String, _ input: String) -> String {
     case "gateway_feedback": return classifyGatewayFeedback(values["hex"] ?? "")
     case "gateway_clipboard":
         if values["file"] != nil { return "invalid:forbidden" }
-        guard ["client_to_provider", "provider_to_client"].contains(values["direction"] ?? ""), values["encoding"] == "utf-8", let token = values["loop_token"], (16...128).contains(token.utf8.count), let text = values["text"], text.utf8.count <= 65536 else { return "invalid:clipboard" }
+        guard let direction = values["direction"], let text = values["text"],
+              let encoding = values["encoding"], let token = values["loop_token"],
+              (try? GatewayClipboardText(
+                  direction: direction, text: text, encoding: encoding, loopToken: token
+              )) != nil else { return "invalid:clipboard" }
         return "valid"
     case "gateway_clipboard_audit":
         if values["text"] != nil { return "invalid:forbidden" }

tools/run_native_conformance.py

@@ -20,7 +20,14 @@ def main() -> int:
         temp = pathlib.Path(directory)
         rust_bin = temp / "rust-conformance"
         swift_bin = temp / "swift-conformance"
-        run(["rustc", "tools/native_conformance.rs", "-O", "-o", str(rust_bin)])
+        rust_source = temp / "main.rs"
+        rust_source.write_text(
+            (ROOT / "gen/rust/protocol.rs").read_text(encoding="utf-8")
+            + "\n"
+            + (ROOT / "tools/native_conformance.rs").read_text(encoding="utf-8"),
+            encoding="utf-8",
+        )
+        run(["rustc", str(rust_source), "-O", "-o", str(rust_bin)])
         run([str(rust_bin)])
         main_source = temp / "main.swift"
         main_source.write_text((ROOT / "tools/native_conformance.swift").read_text(encoding="utf-8"), encoding="utf-8")

tools/test_check_scope.py

@@ -0,0 +1,100 @@
+#!/usr/bin/env python3
+"""Focused contract-boundary regressions for check_scope.py."""
+
+from __future__ import annotations
+
+import json
+import pathlib
+import tempfile
+
+import check_scope
+
+
+TEXT_PATHS = (
+    "openapi/control-v1.yaml",
+    "proto/versevdi/control/v1/control.proto",
+    "proto/versevdi/tunnel/v1/tunnel.proto",
+    "frames/datagram-v1.md",
+    "frames/registry.json",
+    "registries/features.json",
+    "registries/datagrams.json",
+)
+PROVIDER_WORK_PROTO = """
+message ProviderSessionWork {
+  string client_private_key_pem = 15;
+}
+"""
+
+
+def schema_with_private_key(owner: str) -> dict[str, object]:
+    definitions = {
+        name: {"type": "object", "properties": {}}
+        for name in (
+            "ConnectionManifest",
+            "ManifestGateway",
+            "ManifestTunnel",
+            "ManifestProfile",
+            "ManifestBounds",
+            "GrantReference",
+            "ProviderSessionWork",
+        )
+    }
+    definitions[owner]["properties"] = {
+        "client_private_key_pem": {"type": "string"},
+    }
+    return {"$defs": definitions}
+
+
+def run_scope(schema: dict[str, object], overrides: dict[str, str] | None = None) -> None:
+    with tempfile.TemporaryDirectory() as directory:
+        root = pathlib.Path(directory)
+        schema_path = root / "schemas/control-v1.schema.json"
+        schema_path.parent.mkdir(parents=True)
+        schema_path.write_text(json.dumps(schema), encoding="utf-8")
+        for relative in TEXT_PATHS:
+            path = root / relative
+            path.parent.mkdir(parents=True, exist_ok=True)
+            default = PROVIDER_WORK_PROTO if relative == "proto/versevdi/tunnel/v1/tunnel.proto" else ""
+            path.write_text((overrides or {}).get(relative, default), encoding="utf-8")
+        (root / "gen").mkdir()
+
+        original_root = check_scope.ROOT
+        check_scope.ROOT = root
+        try:
+            check_scope.check_manifest_schema()
+            check_scope.check_text_boundaries()
+        finally:
+            check_scope.ROOT = original_root
+
+
+def expect_rejected(schema: dict[str, object], overrides: dict[str, str] | None = None) -> None:
+    try:
+        run_scope(schema, overrides)
+    except ValueError:
+        return
+    raise AssertionError("client-visible private-key material was accepted")
+
+
+def main() -> None:
+    run_scope(schema_with_private_key("ProviderSessionWork"))
+    expect_rejected(schema_with_private_key("ConnectionManifest"))
+    expect_rejected(schema_with_private_key("ManifestProfile"))
+    expect_rejected(
+        schema_with_private_key("ProviderSessionWork"),
+        {
+            "proto/versevdi/tunnel/v1/tunnel.proto": """
+message ConnectionManifest {
+  string client_private_key_pem = 1;
+}
+"""
+        },
+    )
+    expect_rejected(
+        schema_with_private_key("ProviderSessionWork"),
+        {"frames/datagram-v1.md": "client_private_key_pem"},
+    )
+    print("Protocol contract-aware scope regression passed")
+
+
+if __name__ == "__main__":
+    main()

tools/test_generated_contracts.py

@@ -85,6 +85,25 @@ do {
     )
     fatalError("invalid allocation bounds were accepted")
 } catch { }
+for text in [
+    String(repeating: "a", count: 65536),
+    String(repeating: "é", count: 32768),
+    String(repeating: "\\\"", count: 32768),
+] {
+    let clipboard = try GatewayClipboardText(
+        direction: "client_to_provider", text: text, encoding: "utf-8",
+        loopToken: "abcdefghijklmnop"
+    )
+    let decoded = try GatewayClipboardText.decodeJSON(clipboard.encodeJSON())
+    guard decoded.text == text else { fatalError("clipboard text changed during round-trip") }
+}
+do {
+    _ = try GatewayClipboardText(
+        direction: "client_to_provider", text: String(repeating: "a", count: 65537),
+        encoding: "utf-8", loopToken: "abcdefghijklmnop"
+    )
+    fatalError("oversized clipboard text was accepted")
+} catch { }
 """,
             encoding="utf-8",
         )
@@ -130,6 +149,19 @@ fn main() {
     assert!(AllocationPolicy::new(
         100, 50, 25, "standard".into(), "audience".into(), "verse".into(), 1, 60, 300,
     ).is_err());
+    for text in [
+        "a".repeat(65536),
+        "é".repeat(32768),
+        "\\\"".repeat(32768),
+    ] {
+        assert!(GatewayClipboardText::new(
+            "client_to_provider".into(), text, "utf-8".into(), "abcdefghijklmnop".into(),
+        ).is_ok());
+    }
+    assert!(GatewayClipboardText::new(
+        "client_to_provider".into(), "a".repeat(65537), "utf-8".into(),
+        "abcdefghijklmnop".into(),
+    ).is_err());
 }
 """
             )