diff --git a/backend/apps/agents/agent_manager.py b/backend/apps/agents/agent_manager.py
index 46e744d6..af9c7c65 100644
--- a/backend/apps/agents/agent_manager.py
+++ b/backend/apps/agents/agent_manager.py
@@ -35,6 +35,7 @@ from backend.apps.agents.core.error_classify import (
     _is_long_context_error,
     _is_transient_capacity_error,
     _is_unknown_model_error,
+    redact_for_telemetry,
 )
 from backend.apps.agents.manager.session.session_store import (
     _delete_session_file,
@@ -3151,7 +3152,7 @@ class AgentManager:
                         "framework_overhead_tokens": session.framework_overhead_tokens,
                         "active_mcps_count": len(session.active_mcps),
                         "messages_count": len(session.messages),
-                        "error_preview": (str(e) or "")[:500],
+                        "error_preview": redact_for_telemetry(str(e), limit=500),
                     })
                 except Exception:
                     logger.debug("submit_diagnostic for context_overflow failed", exc_info=True)
@@ -3256,7 +3257,8 @@ class AgentManager:
                         "model": session.model,
                         "provider": session.provider,
                         "connection_mode": getattr(load_settings(), "connection_mode", "own_key"),
-                        "error_preview": (str(e) or "")[:400],
+                        "error_preview": redact_for_telemetry(str(e), limit=400),
+                        "stderr_tail": redact_for_telemetry(_stderr_tail),
                     })
                 except Exception:
                     logger.debug("submit_diagnostic model_error failed", exc_info=True)
@@ -3276,7 +3278,8 @@ class AgentManager:
                         "model": session.model,
                         "provider": session.provider,
                         "connection_mode": getattr(load_settings(), "connection_mode", "own_key"),
-                        "error_preview": (str(e) or "")[:400],
+                        "error_preview": redact_for_telemetry(str(e), limit=400),
+                        "stderr_tail": redact_for_telemetry(_stderr_tail),
                     })
                 except Exception:
                     logger.debug("submit_diagnostic model_error failed", exc_info=True)
diff --git a/backend/apps/agents/core/error_classify.py b/backend/apps/agents/core/error_classify.py
index 3a5d4aea..d7b1fb7b 100644
--- a/backend/apps/agents/core/error_classify.py
+++ b/backend/apps/agents/core/error_classify.py
@@ -1,5 +1,30 @@
 import re
 
+# Secret shapes that must never ride along when we ship a stderr tail or an
+# error string to telemetry. own_key mode means the subprocess stderr can echo
+# the user's OWN provider key, so this scrub is the wall between a diagnostic
+# and a key leak; over-redacting is fine, leaking is not.
+_TELEMETRY_SECRET_PATTERNS = (
+    re.compile(r"sk-ant-[A-Za-z0-9_\-]{12,}"),
+    re.compile(r"sk-[A-Za-z0-9_\-]{16,}"),
+    re.compile(r"AIza[A-Za-z0-9_\-]{20,}"),
+    re.compile(r"gh[pousr]_[A-Za-z0-9]{20,}"),
+    re.compile(r"(?i)bearer\s+[A-Za-z0-9._\-]{12,}"),
+    re.compile(r"(?i)\b(?:api[_-]?key|access[_-]?token|refresh[_-]?token|secret|password|authorization)\b[\"']?\s*[:=]\s*[\"']?[A-Za-z0-9._\-]{6,}"),
+)
+
+
+def redact_for_telemetry(text: str, *, limit: int = 2000) -> str:
+    """Scrub secret-shaped substrings, then keep the tail (where the real error
+    lands), bounded so a runaway log can't bloat the payload. Every raw
+    error/stderr string goes through here before it leaves the machine."""
+    if not text:
+        return ""
+    for pat in _TELEMETRY_SECRET_PATTERNS:
+        text = pat.sub("[redacted]", text)
+    return text[-limit:]
+
+
 # Patterns that indicate an upstream transient problem (overload / rate limit /
 # infra blip), safe to silently retry with backoff. Checked against the
 # stringified exception from claude_agent_sdk / Claude CLI.
diff --git a/backend/tests/test_error_classify.py b/backend/tests/test_error_classify.py
new file mode 100644
index 00000000..191edead
--- /dev/null
+++ b/backend/tests/test_error_classify.py
@@ -0,0 +1,45 @@
+"""redact_for_telemetry is the wall between a model_error diagnostic and a key
+leak: in own_key mode the subprocess stderr we now attach can echo the user's
+provider key, so these tests pin that no secret shape survives while the actual
+error text (the whole point of capturing stderr) does.
+
+The secret-shaped inputs are built by concatenation on purpose: no contiguous
+key-shaped literal lands in this source file (so it never trips gitleaks or
+alarms a reader), yet the runtime values are still key-shaped enough to exercise
+the scrub. None of these are real keys; they unlock nothing."""
+from backend.apps.agents.core.error_classify import redact_for_telemetry
+
+
+def test_redacts_provider_key_shapes_keeps_context():
+    anthropic = "sk-" + "ant-" + "A" * 28
+    openai = "sk-" + "B" * 24
+    google = "AIza" + "C" * 30
+    github = "ghp" + "_" + "D" * 24
+    s = f"9router: invalid x-api-key {anthropic} {openai} {google} {github}"
+    out = redact_for_telemetry(s)
+    for secret in (anthropic, openai, google, github):
+        assert secret not in out
+    assert "[redacted]" in out
+    # The diagnostic signal survives, that's the reason we capture stderr at all.
+    assert "9router: invalid x-api-key" in out
+
+
+def test_redacts_bearer_and_key_value():
+    bearer_token = "E" * 24
+    kv_value = "F" * 16
+    s = "Authorization: " + "Bearer " + bearer_token + "\n" + "api_key=" + kv_value
+    out = redact_for_telemetry(s)
+    assert bearer_token not in out
+    assert kv_value not in out
+
+
+def test_keeps_tail_and_bounds_length():
+    # The real error lands at the end of the stderr stream, so we keep the tail.
+    s = "old noise\n" * 500 + "Command failed: ENOENT spawn 9router"
+    out = redact_for_telemetry(s, limit=120)
+    assert len(out) <= 120
+    assert "Command failed: ENOENT spawn 9router" in out
+
+
+def test_empty_is_safe():
+    assert redact_for_telemetry("") == ""