fix(preflight): timeout hung adapter auth checks (OPS-002)

scripts/lib/adapter_preflight.py

@@ -18,6 +18,25 @@
     re.MULTILINE | re.DOTALL,
 )
 
+# Auth probes must not hang forever (OPS-002 / PREFLIGHT-01). Override via
+# FLEET_ADAPTER_AUTH_TIMEOUT_S (seconds); non-positive or invalid values fall
+# back to the default.
+DEFAULT_AUTH_TIMEOUT_S = 30.0
+_AUTH_TIMEOUT_ENV = "FLEET_ADAPTER_AUTH_TIMEOUT_S"
+
+
+def _auth_timeout_s(environ: Mapping[str, str] = os.environ) -> float:
+    raw = environ.get(_AUTH_TIMEOUT_ENV)
+    if raw is None or raw == "":
+        return DEFAULT_AUTH_TIMEOUT_S
+    try:
+        value = float(raw)
+    except ValueError:
+        return DEFAULT_AUTH_TIMEOUT_S
+    if value <= 0:
+        return DEFAULT_AUTH_TIMEOUT_S
+    return value
+
 
 @dataclass(frozen=True)
 class Intent:
@@ -124,7 +143,20 @@ def check(
             continue
         if entry.get("skip_if_intent") == intent_name:
             continue
-        result = run(shlex.split(command), capture_output=True, text=True, check=False)
+        timeout_s = _auth_timeout_s(environ)
+        try:
+            result = run(
+                shlex.split(command),
+                capture_output=True,
+                text=True,
+                check=False,
+                timeout=timeout_s,
+            )
+        except subprocess.TimeoutExpired:
+            failures.append(
+                f"auth check timed out after {timeout_s:g}s: {command}"
+            )
+            continue
         if result.returncode != 0:
             failures.append(f"auth check failed ({result.returncode}): {command}")
 

skills/autonomous-fleet-core/assets/substrate/lib/adapter_preflight.py

@@ -18,6 +18,25 @@
     re.MULTILINE | re.DOTALL,
 )
 
+# Auth probes must not hang forever (OPS-002 / PREFLIGHT-01). Override via
+# FLEET_ADAPTER_AUTH_TIMEOUT_S (seconds); non-positive or invalid values fall
+# back to the default.
+DEFAULT_AUTH_TIMEOUT_S = 30.0
+_AUTH_TIMEOUT_ENV = "FLEET_ADAPTER_AUTH_TIMEOUT_S"
+
+
+def _auth_timeout_s(environ: Mapping[str, str] = os.environ) -> float:
+    raw = environ.get(_AUTH_TIMEOUT_ENV)
+    if raw is None or raw == "":
+        return DEFAULT_AUTH_TIMEOUT_S
+    try:
+        value = float(raw)
+    except ValueError:
+        return DEFAULT_AUTH_TIMEOUT_S
+    if value <= 0:
+        return DEFAULT_AUTH_TIMEOUT_S
+    return value
+
 
 @dataclass(frozen=True)
 class Intent:
@@ -124,7 +143,20 @@ def check(
             continue
         if entry.get("skip_if_intent") == intent_name:
             continue
-        result = run(shlex.split(command), capture_output=True, text=True, check=False)
+        timeout_s = _auth_timeout_s(environ)
+        try:
+            result = run(
+                shlex.split(command),
+                capture_output=True,
+                text=True,
+                check=False,
+                timeout=timeout_s,
+            )
+        except subprocess.TimeoutExpired:
+            failures.append(
+                f"auth check timed out after {timeout_s:g}s: {command}"
+            )
+            continue
         if result.returncode != 0:
             failures.append(f"auth check failed ({result.returncode}): {command}")
 

skills/autonomous-fleet-core/assets/substrate/substrate-manifest.json

@@ -3,7 +3,7 @@
   "files": {
     "emit_trace.py": "4425a4dd35f40b230eb566330bb2c28fd487892a717e440947c31327388805ad",
     "lib/__init__.py": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
-    "lib/adapter_preflight.py": "c0c5a7a57840374b92cca95e7ea043ecfcd499ab2f27d01cedc53f0ed72b8ed5",
+    "lib/adapter_preflight.py": "4cfd48fb36ce2c0216c9d5112d76864c5710dcafe561e7942ae1aa059c150b24",
     "lib/analyze_cost.py": "e0359e9eabfc0c01dd5241bce8c1c124698da3a62a9fee9fbfcaa023a849b645",
     "lib/analyze_seat.py": "ee0faf086e679fd50315f2f4b0293590e8e73a0961fcad9388899902700d1c67",
     "lib/community_preflight.py": "94a6eb2d68397db7c0f8dad36d5ce88b40893f12fd40765a97d9aab0946f2d4c",

tests/test_adapter_preflight.py

@@ -1,17 +1,23 @@
 """Tests for adapter requires-block preflight checks."""
 from __future__ import annotations
 
+import subprocess
+import sys
 from pathlib import Path
 from types import SimpleNamespace
 
 import pytest
 
-import sys
-
 ROOT = Path(__file__).resolve().parents[1]
 sys.path.insert(0, str(ROOT / "scripts"))
 
-from lib.adapter_preflight import Intent, activity_hooks_advisory, check, load_requires  # noqa: E402
+from lib.adapter_preflight import (  # noqa: E402
+    DEFAULT_AUTH_TIMEOUT_S,
+    Intent,
+    activity_hooks_advisory,
+    check,
+    load_requires,
+)
 
 
 def _which_factory(found: set[str]):
@@ -21,15 +27,29 @@ def which(binary: str) -> str | None:
     return which
 
 
-def _runner(returncode: int, calls: list[list[str]]):
+def _runner(returncode: int, calls: list[list[str]], *, timeout_s: float = DEFAULT_AUTH_TIMEOUT_S):
     def run(args: list[str], **kwargs):
         calls.append(args)
-        assert kwargs == {"capture_output": True, "text": True, "check": False}
+        assert kwargs == {
+            "capture_output": True,
+            "text": True,
+            "check": False,
+            "timeout": timeout_s,
+        }
         return SimpleNamespace(returncode=returncode)
 
     return run
 
 
+def _timeout_runner(calls: list[list[str]], *, timeout_s: float = DEFAULT_AUTH_TIMEOUT_S):
+    def run(args: list[str], **kwargs):
+        calls.append(args)
+        assert kwargs["timeout"] == timeout_s
+        raise subprocess.TimeoutExpired(cmd=args, timeout=timeout_s)
+
+    return run
+
+
 def test_wiring_only_skips_all_checks() -> None:
     failures = check(
         {
@@ -151,6 +171,50 @@ def test_auth_entry_with_empty_check_reports_failure() -> None:
     assert calls == []
 
 
+def test_auth_timeout_reports_clear_failure_via_injected_run() -> None:
+    calls: list[list[str]] = []
+    failures = check(
+        {"bins": [], "env": [], "auth": [{"check": "gh auth status"}]},
+        Intent(scm=True),
+        which=_which_factory(set()),
+        run=_timeout_runner(calls),
+        environ={},
+    )
+
+    assert failures == [
+        f"auth check timed out after {DEFAULT_AUTH_TIMEOUT_S:g}s: gh auth status"
+    ]
+    assert calls == [["gh", "auth", "status"]]
+
+
+def test_auth_timeout_honors_fleet_adapter_auth_timeout_env() -> None:
+    calls: list[list[str]] = []
+    failures = check(
+        {"bins": [], "env": [], "auth": [{"check": "gh auth status"}]},
+        Intent(scm=True),
+        which=_which_factory(set()),
+        run=_timeout_runner(calls, timeout_s=12.5),
+        environ={"FLEET_ADAPTER_AUTH_TIMEOUT_S": "12.5"},
+    )
+
+    assert failures == ["auth check timed out after 12.5s: gh auth status"]
+    assert calls == [["gh", "auth", "status"]]
+
+
+def test_auth_timeout_invalid_env_falls_back_to_default() -> None:
+    calls: list[list[str]] = []
+    failures = check(
+        {"bins": [], "env": [], "auth": [{"check": "gh auth status"}]},
+        Intent(scm=True),
+        which=_which_factory({"gh"}),
+        run=_runner(0, calls),
+        environ={"FLEET_ADAPTER_AUTH_TIMEOUT_S": "nope"},
+    )
+
+    assert failures == []
+    assert calls == [["gh", "auth", "status"]]
+
+
 def test_load_requires_reads_fenced_yaml_requires_block(tmp_path: Path) -> None:
     adapter = tmp_path / "adapter"
     adapter.mkdir()