feat(redact): cover UUIDs and blockchain credentials

Cargo.toml

@@ -168,6 +168,8 @@ arc-swap = "1"
 blake3 = "1"
 # PKCE 的 S256。版本跟着 reqwest 的传递依赖走，不引新的树
 sha2 = "0.11"
+bech32 = "0.9.1"
+curve25519-dalek = { version = "4.1.3", default-features = false, features = ["alloc", "precomputed-tables"] }
 uuid = { version = "1", features = ["v4", "serde"] }
 rand = "0.10"
 # 发给 Codex 后端的请求体用 zstd 压缩（Codex 直连它时就这么发）。只要流式编码，

crates/tw-guard/Cargo.toml

@@ -20,6 +20,8 @@ regex-syntax = { workspace = true }
 memchr = { workspace = true }
 aho-corasick = { workspace = true }
 sha2 = { workspace = true }
+bech32 = { workspace = true }
+curve25519-dalek = { workspace = true }
 thiserror = { workspace = true }
 serde_yaml_ng = { workspace = true }
 tw-dialect = { workspace = true }

crates/tw-guard/data/redact.yaml

@@ -44,8 +44,8 @@
 #   **前缀类的规则不写 `about`**：说明由 build.rs 按前缀和长度生成，和判据不会说两样。
 #   加一条前缀类的规则只改这个文件。
 #
-# 每一条内置规则都要求「前缀明确」或者「结构上无法误认」。代码里的 hash、base64 的图片、
-# UUID 全都长得像「一长串随机字符」，按那个判据匹配会疯狂误报。
+# 凭据规则按前缀、结构或字段位置识别。UUID、裸十六进制和链上地址另作隐私规则，
+# 不把这些无从区分用途的格式单独当成密钥外传证据。
 
 kinds:
   - slug: saved
@@ -869,8 +869,87 @@ rules:
     on_by_default: true
     name: Extended private key
     name_zh: 扩展私钥
-    about: 'Starts with `xprv`, `tprv`, `yprv` or `zprv` and passes Base58Check'
-    about_zh: '`xprv`、`tprv`、`yprv`、`zprv` 开头，Base58Check 校验通过'
+    about: 'BIP32 / SLIP-0132 extended private keys: `xprv`, `tprv`, `yprv`, `Yprv`, `zprv`, `Zprv`, `uprv`, `Uprv`, `vprv` or `Vprv`; valid Base58Check, private marker and secp256k1 scalar'
+    about_zh: 'BIP32 / SLIP-0132 扩展私钥：`xprv`、`tprv`、`yprv`、`Yprv`、`zprv`、`Zprv`、`uprv`、`Uprv`、`vprv`、`Vprv`；Base58Check、私钥标记及 secp256k1 标量均有效'
+
+
+  - id: uuid
+    kind: device
+    on_by_default: true
+    name: UUID / compact identifier
+    name_zh: UUID／紧凑标识符
+    about: 'Hyphenated UUIDs and 32 hexadecimal characters, ignoring case and without version restrictions, including order and trace IDs; compact hashes of the same shape are also replaced'
+    about_zh: '带连字符的 UUID 和 32 位十六进制串，不区分大小写、不限制版本，包括订单号和追踪 ID；同形状的紧凑哈希也替换'
+  - id: crypto-hex-key
+    kind: crypto
+    on_by_default: true
+    name: Raw wallet key / 32-byte hash
+    name_zh: 裸钱包私钥／32 字节哈希
+    about: '64 hexadecimal characters with an optional `0x` prefix; covers EVM and TRON private keys as well as indistinguishable transaction hashes and digests'
+    about_zh: '64 位十六进制串，可带 `0x` 前缀；覆盖 EVM、TRON 裸私钥以及格式无法区分的交易哈希和摘要'
+  - id: crypto-solana-key
+    kind: crypto
+    on_by_default: true
+    name: Solana keypair
+    name_zh: Solana 密钥对
+    about: '64-byte Ed25519 seed and public key encoded as Base58 or a pasted JSON byte array; the public half must match the seed, excluding signatures of the same length'
+    about_zh: 'Base58 编码或作为文本粘贴的 JSON 字节数组，含 64 字节 Ed25519 种子与公钥；由种子派生的公钥须一致，排除同长度签名'
+  - id: crypto-sui-key
+    kind: crypto
+    on_by_default: true
+    name: Sui private key
+    name_zh: Sui 私钥
+    about: '`suiprivkey1` Bech32 encoding with a valid checksum, a supported signature scheme and 32 key bytes'
+    about_zh: '`suiprivkey1` Bech32 编码，校验和有效，含支持的签名方案和 32 字节密钥'
+  - id: crypto-stellar-key
+    kind: crypto
+    on_by_default: true
+    name: Stellar secret seed
+    name_zh: Stellar 私密种子
+    about: 'Stellar `S` StrKey seed with the correct version, 32-byte payload and CRC16-XModem checksum'
+    about_zh: 'Stellar `S` StrKey 种子，版本、32 字节载荷和 CRC16-XModem 校验均有效'
+  - id: crypto-xrp-seed
+    kind: crypto
+    on_by_default: true
+    name: XRP family seed
+    name_zh: XRP 账户种子
+    about: 'XRP-alphabet Base58Check family seeds, including secp256k1 and Ed25519 version encodings'
+    about_zh: 'XRP 字母表的 Base58Check 账户种子，包括 secp256k1 和 Ed25519 版本'
+  - id: crypto-address
+    kind: financial
+    on_by_default: true
+    name: Blockchain account / extended public key
+    name_zh: 区块链账户／扩展公钥
+    about: 'EVM hex addresses, checksummed Bitcoin/TRON/XRP/Stellar addresses, Bitcoin witness addresses and BIP32 / SLIP-0132 extended public keys; also 32-byte Base58 account shapes such as Solana, which lack a checksum'
+    about_zh: 'EVM 十六进制地址，经校验的 Bitcoin／TRON／XRP／Stellar 地址、Bitcoin 隔离见证地址和 BIP32／SLIP-0132 扩展公钥；也覆盖 Solana 等无校验和的 32 字节 Base58 账户格式'
+  - id: crypto-secret-field
+    kind: in-context
+    on_by_default: true
+    name: Wallet secret in a field
+    name_zh: 字段中的钱包机密
+    about: 'Literal private keys, seeds, mnemonic/recovery phrases, passphrases, recovery codes and MPC/backup shares in named fields, including `pk`, `privateKey`, `PRIVY_WALLET_PRIVATE_KEY` and Chinese labels; only the value is replaced, with no guessed key length'
+    about_zh: '按字段识别私钥、种子、助记词、恢复短语、口令、恢复码和 MPC／备份份额，包括 `pk`、`privateKey`、`PRIVY_WALLET_PRIVATE_KEY` 和中文标签；只换值，不猜测密钥长度'
+  - id: exchange-credential
+    kind: in-context
+    on_by_default: true
+    name: Trading API / signed authentication credential
+    name_zh: 交易 API／签名鉴权凭证
+    about: 'Literal API keys, secrets, passphrases, signing memos, OTPs and authentication signatures in SDK/config fields or trading headers; Binance, OKX, Bybit, Coinbase, Kraken, KuCoin, Bitget, Gate, MEXC, Gemini, Bitfinex, HTX, BitMart, Phemex, Bitstamp, Polymarket and predict.fun; only the value is replaced'
+    about_zh: 'SDK／配置字段或交易请求头中的 API key、secret、passphrase、签名 Memo、OTP 和鉴权签名；覆盖 Binance、OKX、Bybit、Coinbase、Kraken、KuCoin、Bitget、Gate、MEXC、Gemini、Bitfinex、HTX、BitMart、Phemex、Bitstamp、Polymarket 和 predict.fun；只换值'
+  - id: crypto-keystore
+    kind: in-context
+    on_by_default: true
+    name: Wallet keystore / ciphertext
+    name_zh: 钱包密钥库／密文
+    about: 'Literal `keystore` or `ciphertext` fields, including Ethereum V3 encrypted key material; pasted JSON objects are protected as text and only the value is replaced'
+    about_zh: '字面量 `keystore` 或 `ciphertext` 字段，包括 Ethereum V3 加密密钥材料；作为文本粘贴的 JSON 对象也保护，只换值'
+  - id: crypto-rpc-key
+    kind: in-context
+    on_by_default: true
+    name: Blockchain RPC URL credential
+    name_zh: 区块链 RPC 地址凭证
+    about: 'API credentials in Infura `/v3/`, Alchemy `/v2/`, QuickNode token paths and Helius `api-key` URLs; only the credential part is replaced'
+    about_zh: 'Infura `/v3/`、Alchemy `/v2/`、QuickNode 令牌路径及 Helius `api-key` URL 中的 API 凭据；只换凭据部分'
 
   # ---- 扩充的个人信息规则（判据在 src/redact/personal/）
   - id: cn-resident-id-15

crates/tw-guard/src/redact/engine.rs

@@ -748,14 +748,15 @@ pub(super) fn scan_tokens(
     let mut personal = super::personal::Want::new(set);
     let on = enabled(set, &secrets, &personal);
     if on == 0 {
+        super::secrets::blockchain::scan(text, set, json, None, out);
         personal.text(text, json, None, out);
         return None;
     }
     let pre = prescan(
         text,
         json,
         on,
-        json && (personal.labeled_on() || set.has_custom()),
+        json && (personal.labeled_on() || set.has_custom() || super::secrets::blockchain::on(set)),
     );
     let spans = (json && !pre.quotes.is_empty()).then(|| pre.string_spans(text.len()));
     // 整段文本过一遍的（中文助记词；护照机读区、住址……）在按 token 认的前面，和老写法一样
@@ -837,6 +838,7 @@ pub(super) fn scan_tokens(
         secrets.word_run(text, r.clone(), out);
     }
     spaced.finish(text, out);
+    super::secrets::blockchain::scan(text, set, json, spans.as_deref(), out);
     secrets.finish(text, set, out);
     personal.finish(text, out);
     spans

crates/tw-guard/src/redact/flow.rs

@@ -621,6 +621,39 @@ mod tests {
         assert!(ledger.is_empty());
     }
 
+    #[test]
+    fn uuid_redaction_observes_or_replaces_written_ids_and_restores_them() {
+        let value = "7b1e2d60-6f9a-4dd7-8f3e-21b8f94c77a2";
+        let input = serde_json::json!({
+            "id": value,
+            "model": value,
+            "messages": [{"role": "user", "content": format!("order {value}; key {value}")}]
+        })
+        .to_string();
+        let rules = RuleSet::defaults();
+        for mode in [Mode::Observe, Mode::Enforce] {
+            let (found, ledger) = look(mode, &rules, input.as_bytes(), S);
+            assert_eq!(found.len(), 1);
+            assert_eq!(found[0].rule.id(), "uuid");
+            let (out, ledger) = replace(mode, &rules, input.clone().into(), &ledger, S);
+            let out = String::from_utf8(out.to_vec()).unwrap();
+            if mode == Mode::Observe {
+                assert_eq!(out, input);
+                assert!(ledger.is_empty());
+            } else {
+                let parsed: serde_json::Value = serde_json::from_str(&out).unwrap();
+                assert_eq!(parsed["id"], value);
+                assert_eq!(parsed["model"], value);
+                assert_eq!(
+                    parsed["messages"][0]["content"],
+                    "order <<TW_SECRET_1>>; key <<TW_SECRET_1>>"
+                );
+                assert_eq!(ledger.len(), 1);
+                assert_eq!(crate::redact::replace::restore(&out, &ledger), input);
+            }
+        }
+    }
+
     #[test]
     fn off_does_not_even_look() {
         assert!(find(Mode::Off, &RuleSet::defaults(), &body(), S).is_empty());

crates/tw-guard/src/redact/rules.rs

@@ -335,8 +335,7 @@ pub const CARD_NETWORKS: &[CardNetwork] = &[
 /// 全部内置规则，从 `data/redact.yaml` 编译期生成（见 build.rs）。界面上先按 [`Kind`] 的
 /// 顺序分组，组内按这里的顺序（见 `view::redact`）；前缀表也按这里的顺序比。
 ///
-/// API 密钥那一段**只收前缀明确、长度有下限的**。代码里的 hash、base64 的
-/// 图片、UUID 全都长得像「一长串随机字符」，按那个判据匹配会疯狂误报。
+/// API 密钥按前缀、结构或上下文识别；UUID、链上地址和裸十六进制另作隐私规则。
 pub const BUILTINS: &[Builtin] = include!(concat!(env!("OUT_DIR"), "/redact_builtins.rs"));
 
 /// 一个分组叫什么（`data/redact.yaml` 的 `kinds`）。
@@ -410,6 +409,17 @@ const DETECTORS: &[(&str, Matcher)] = &[
     ("crypto-mnemonic", Matcher::Code),
     ("crypto-wif-key", Matcher::Code),
     ("crypto-extended-key", Matcher::Code),
+    ("uuid", Matcher::Code),
+    ("crypto-hex-key", Matcher::Code),
+    ("crypto-solana-key", Matcher::Code),
+    ("crypto-sui-key", Matcher::Code),
+    ("crypto-stellar-key", Matcher::Code),
+    ("crypto-xrp-seed", Matcher::Code),
+    ("crypto-address", Matcher::Code),
+    ("crypto-secret-field", Matcher::Code),
+    ("exchange-credential", Matcher::Code),
+    ("crypto-keystore", Matcher::Code),
+    ("crypto-rpc-key", Matcher::Code),
     // 扩充的个人信息规则（personal/）
     ("cn-resident-id-15", Matcher::Code),
     ("passport-mrz", Matcher::Code),
@@ -2398,6 +2408,16 @@ pub fn masked(rule: &Rule, value: &str) -> String {
     {
         return mask_email(value);
     }
+    if matches!(
+        rule.id(),
+        "uuid"
+            | "crypto-secret-field"
+            | "exchange-credential"
+            | "crypto-keystore"
+            | "crypto-rpc-key"
+    ) {
+        return "…".to_string();
+    }
     match rule.kind() {
         Kind::Internal => value.to_string(),
         k if k.is_personal() => last_four(value),
@@ -2406,7 +2426,7 @@ pub fn masked(rule: &Rule, value: &str) -> String {
         // 按位置认的多半是口令，没有前缀：留头五个就是留了大半个口令。只留最后四位，短的全打掉
         Kind::InContext => masked_unknown(value),
         // 助记词的头几个字母就是头一个词
-        Kind::Crypto if rule.id() == "crypto-mnemonic" => "…".to_string(),
+        Kind::Crypto => "…".to_string(),
         _ => mask(value),
     }
 }
@@ -2658,8 +2678,6 @@ mod tests {
         // **宁可漏，不可吵。**一个天天误报的安全功能，用户第二天就关了。
         for t in [
             "把 sk-test 换成你自己的 key",
-            "const hash = 'a3f5c9e1b7d2f4a6c8e0b2d4f6a8c0e2';",
-            "UUID 是 550e8400-e29b-41d4-a716-446655440000",
             "版本 1.2.3，路径 a.b.c",
             "data:image/png;base64,iVBORw0KGgoAAAANSUhEUg==",
             // 全字母或全数字的一长串是 hash / id，不是 key
@@ -3154,6 +3172,10 @@ mod tests {
                 }
                 Matcher::Code if b.kind.is_personal() => {
                     let text = super::super::personal::tests::sample(b.id)
+                        .or_else(|| {
+                            super::super::secrets::tests::samples::sample(b.id)
+                                .map(|s| s.hits[0].0.clone())
+                        })
                         .unwrap_or_else(|| panic!("{} 没有样本", b.id));
                     let hits = scan_plain(&text, &only);
                     assert!(!hits.is_empty(), "{} 认不出自己的样本：{text}", b.id);

crates/tw-guard/src/redact/rules/oracle.rs

@@ -86,6 +86,7 @@ pub(crate) fn scan_unresolved(text: &str, set: &RuleSet) -> Vec<Hit> {
         }
     });
     spaced.finish(text, &mut out);
+    super::super::secrets::blockchain::scan(text, set, json, None, &mut out);
     secrets.finish(text, set, &mut out);
     personal.finish(text, &mut out);
     custom_hits(text, set, json, &mut out);