A CLI tool that tracks your Claude Agent SDK credit burn rate in real time and routes tasks to the cheapest qualifying model before credits run out
Starting June 15 2026, Anthropic splits Claude subscription billing into two pools. Agent SDK usage (OpenClaw, claude -p, GitHub Actions, third-party agents) gets a separate monthly credit with hard caps ($20 for Pro, $100 for Max 5x, $200 for Max 20x) and no rollover. Developers running agents through Claude subscriptions face a new constraint: fixed monthly budgets that burn fast on inefficient prompts. This tool monitors Agent SDK credit consumption per task, alerts before exhaustion, and optionally reroutes low-priority agent tasks to cheaper providers (OpenRouter, local models) when credits drop below a configurable threshold.
Social Proof 3 sources
Gap Assessment
3 tools exist (OpenRouter, Helicone, LiteLLM) but gaps remain: No awareness of Anthropic Agent SDK credit pools, no automatic fallback when credits run low, no per-task attribution; No Agent SDK credit pool monitoring, no automatic model downgrade routing, no credit exhaustion alerts.
Features3 agent-ready prompts
Competitive LandscapeFREE
| Product | Does | Missing |
|---|---|---|
| OpenRouter | Universal API gateway routing to 400+ models with per-request cost tracking | No awareness of Anthropic Agent SDK credit pools, no automatic fallback when credits run low, no per-task attribution |
| Helicone | LLM observability platform with cost tracking, latency monitoring, and request logging | No Agent SDK credit pool monitoring, no automatic model downgrade routing, no credit exhaustion alerts |
| LiteLLM | Universal LLM proxy supporting 100+ providers with spend tracking and budget limits | No Anthropic Agent SDK credit-specific tracking, budget limits are static not credit-pool-aware |
Leads35BUILDER
Sign in to unlock full access.