From dd9e7f4c33b630234b24ae167b6ae713603beca6 Mon Sep 17 00:00:00 2001
From: Dev Chiniwala <dev.chiniwala@gmail.com>
Date: Thu, 17 Sep 2026 16:17:16 +0530
Subject: [PATCH] fix: redact secrets from audit errors

---
 datamind/capabilities/hooks/audit.py | 21 +++++++++++++++++----
 datamind/tests/test_hooks.py         | 20 ++++++++++++++++++++
 2 files changed, 37 insertions(+), 4 deletions(-)

diff --git a/datamind/capabilities/hooks/audit.py b/datamind/capabilities/hooks/audit.py
index b1b69dd..e22b622 100644
--- a/datamind/capabilities/hooks/audit.py
+++ b/datamind/capabilities/hooks/audit.py
@@ -16,9 +16,9 @@
 
 Secret redaction: arg values whose KEY matches a redaction regex
 (`api_key`, `password`, `token`, `secret`, ...) are replaced with
-`"[REDACTED]"` before logging. Values are not scanned (we trust the
-caller to put secrets in well-named fields, which is true across
-DataMind's own tools).
+`"[REDACTED]"` before logging. Credential-shaped values in exception
+text are redacted as well because provider and transport errors commonly
+include the rejected request detail.
 
 Concurrency: writes are serialised via an asyncio.Lock per AuditLogHook
 instance. For multi-process deployments use one process per profile
@@ -45,6 +45,11 @@
     r"(?i)(api[_-]?key|password|passwd|token|secret|authorization|bearer|access[_-]?key|client[_-]?secret)"
 )
 _REDACTED = "[REDACTED]"
+_BEARER_TEXT_RE = re.compile(r"(?i)\bBearer\s+([^\s,;]+)")
+_SECRET_TEXT_RE = re.compile(
+    r"(?i)\b(api[_-]?key|password|passwd|token|secret|authorization|access[_-]?key|client[_-]?secret)\b"
+    r"(\s*[:=]\s*)([^\s,;]+)"
+)
 
 _HASH_HEX_LEN = 16  # truncated SHA-256 hex; 64 bits of collision resistance
 
@@ -74,6 +79,14 @@ def _redact(value: Any) -> Any:
     return value
 
 
+def _redact_text(value: str) -> str:
+    """Redact common credential formats embedded in diagnostic text."""
+    value = _BEARER_TEXT_RE.sub(f"Bearer {_REDACTED}", value)
+    return _SECRET_TEXT_RE.sub(
+        lambda match: f"{match.group(1)}{match.group(2)}{_REDACTED}", value
+    )
+
+
 def _decision_to_record(decision: HookDecision) -> dict[str, Any]:
     if isinstance(decision, Allow):
         return {"kind": "allow"}
@@ -163,7 +176,7 @@ async def post_tool_use(
             "error": (
                 None
                 if error is None
-                else f"{type(error).__name__}: {error}"
+                else _redact_text(f"{type(error).__name__}: {error}")
             ),
         }
         await self._append(record)
diff --git a/datamind/tests/test_hooks.py b/datamind/tests/test_hooks.py
index 55b5567..cf8e2e0 100644
--- a/datamind/tests/test_hooks.py
+++ b/datamind/tests/test_hooks.py
@@ -322,6 +322,26 @@ async def test_audit_log_redacts_secret_keys(tmp_path: Path):
     assert rec["args"]["username"] == "ann"
 
 
+@pytest.mark.asyncio
+async def test_audit_log_redacts_secrets_in_errors(tmp_path: Path):
+    audit = tmp_path / "audit.jsonl"
+    hook = AuditLogHook(audit_path=audit)
+    await hook.post_tool_use(
+        _ctx(),
+        "fake_tool",
+        {},
+        result=None,
+        error=RuntimeError(
+            "request failed: api_key=sk-live-secret Authorization: Bearer bearer-secret"
+        ),
+    )
+
+    rec = json.loads(audit.read_text().splitlines()[-1])
+    assert "sk-live-secret" not in rec["error"]
+    assert "bearer-secret" not in rec["error"]
+    assert "[REDACTED]" in rec["error"]
+
+
 @pytest.mark.asyncio
 async def test_audit_log_records_denied_decision(tmp_path: Path):
     audit = tmp_path / "audit.jsonl"