Claude adds Agent SDK credits
Programmatic agent usage gets a separate monthly credit pool.
Claude is separating agent-style usage from ordinary chat limits, which helps small automations but still requires buyers to watch monthly credit caps.
What changed
Anthropic says that starting June 15, 2026, Claude Agent SDK and non-interactive claude -p usage will no longer count against regular Claude subscription usage limits. Instead, eligible Pro, Max, Team, and Enterprise users can claim a separate monthly Agent SDK credit.
The published monthly credits include $20 for Pro, $100 for Max 5x, and $200 for Max 20x. Anthropic also says the credit is per user, refreshes monthly, and unused credit does not roll over.
Why it matters for buyers
This is good for people testing small agent workflows, but it is not the same as unlimited API access. The credit applies to Agent SDK usage, claude -p, Claude Code GitHub Actions, and supported third-party apps that authenticate through the Agent SDK. Interactive Claude chat and Claude Code in the terminal continue to use normal subscription limits.
Small scripts, test agents, low-volume automations.
Credits are per user and do not pool across teammates.
If extra usage is off, requests stop when the credit runs out.
Aitoque take
For buyers, the key question is whether you need Claude for chat, coding in the IDE, or automated agent runs. These now map to different limit pools. Choose the plan around the workflow, not only the model name.