Skip to content

Compress tool outputs, logs, files, and RAG chunks before they reach the LLM. 20% fewer tokens for coding agents, 60-95% fewer tokens for JSON, same answers. Library, proxy, MCP server.

fix(subscription): raise default OAuth usage poll interval to 300s

chopratejasa1d9832e27dd
diff --git a/headroom/cli/proxy.py b/headroom/cli/proxy.py
index 8a222178f..738d7d8e4 100644
--- a/headroom/cli/proxy.py
+++ b/headroom/cli/proxy.py
@@ -57,6 +57,26 @@ from .main import main
@click.option("--no-optimize", is_flag=True, help="Disable optimization (passthrough mode)")
@click.option("--no-cache", is_flag=True, help="Disable semantic caching")
@click.option("--no-rate-limit", is_flag=True, help="Disable rate limiting")
+ "--no-subscription-tracking",
+ is_flag=True,
+ envvar="HEADROOM_NO_SUBSCRIPTION_TRACKING",
+ help=(
+ "Disable the Anthropic Claude Code subscription usage poller "
+ "(GET /api/oauth/usage). Env: HEADROOM_NO_SUBSCRIPTION_TRACKING."
+ ),
+)
+ "--subscription-poll-interval",
+ type=int,
+ default=None,
+ envvar="HEADROOM_SUBSCRIPTION_POLL_INTERVAL",
+ help=(
+ "Seconds between Anthropic subscription usage polls (1–3600, default 300). "
+ "Lower values give fresher /stats but risk 429s from Anthropic. "
+ "Env: HEADROOM_SUBSCRIPTION_POLL_INTERVAL."
+ ),
+)
@click.option(
"--retry-max-attempts",
type=int,
@@ -254,6 +274,8 @@ def proxy(
no_optimize: bool,
no_cache: bool,
no_rate_limit: bool,
+ no_subscription_tracking: bool,
+ subscription_poll_interval: int | None,
retry_max_attempts: int | None,
connect_timeout_seconds: int | None,
anthropic_pre_upstream_concurrency: int | None,
@@ -381,6 +403,10 @@ def proxy(
optimize=not no_optimize,
cache_enabled=not no_cache,
rate_limit_enabled=not no_rate_limit,
+ subscription_tracking_enabled=not no_subscription_tracking,
+ subscription_poll_interval_s=(
+ subscription_poll_interval if subscription_poll_interval is not None else 300
+ ),
retry_max_attempts=retry_max_attempts if retry_max_attempts is not None else 3,
connect_timeout_seconds=connect_timeout_seconds
if connect_timeout_seconds is not None
diff --git a/headroom/proxy/models.py b/headroom/proxy/models.py
index 1838d015b..1bbf5f3fd 100644
--- a/headroom/proxy/models.py
+++ b/headroom/proxy/models.py
@@ -218,7 +218,7 @@ class ProxyConfig:
# Subscription Window Tracking (Anthropic OAuth accounts)
subscription_tracking_enabled: bool = True
- subscription_poll_interval_s: int = 10
+ subscription_poll_interval_s: int = 300
subscription_active_window_s: int = 60
# Stateless mode — disable all filesystem writes for read-only / container deployments
diff --git a/headroom/subscription/tracker.py b/headroom/subscription/tracker.py
index 602cee0dc..70e5e9467 100644
--- a/headroom/subscription/tracker.py
+++ b/headroom/subscription/tracker.py
@@ -43,7 +43,7 @@ from headroom.subscription.models import (
logger = logging.getLogger(__name__)
-_DEFAULT_POLL_INTERVAL_S = 10
+_DEFAULT_POLL_INTERVAL_S = 300
_DEFAULT_ACTIVE_WINDOW_S = 60
_PERSIST_FILE_ENV = _paths.HEADROOM_SUBSCRIPTION_STATE_PATH_ENV
_DEFAULT_PERSIST_DIR = ".headroom"
@@ -70,7 +70,7 @@ class SubscriptionTracker(QuotaTracker):
alongside the Codex and Copilot trackers.
Args:
- poll_interval_s: Seconds between polls while active (1–300).
+ poll_interval_s: Seconds between polls while active (1–3600, default 300).
active_window_s: Seconds since last notify_active call that keeps
polling alive (default 60 = 1 minute).
enabled: Set to ``False`` to disable tracking (mirrors
@@ -92,7 +92,7 @@ class SubscriptionTracker(QuotaTracker):
client: SubscriptionClient | None = None,
) -> None:
self._enabled = enabled
- self._poll_interval_s = max(1, min(poll_interval_s, 300))
+ self._poll_interval_s = max(1, min(poll_interval_s, 3600))
self._active_window_s = max(5.0, active_window_s)
self._persist_path = persist_path or _get_persist_path()
self._client = client or SubscriptionClient()