Claude Usage Limits
Lets your agent check how much of your Claude Pro or Max limit is left (5-hour session and weekly) and when each resets, so it sizes heavy work to what remains.
Free
Works with the AI Agents you already use
Claude Usage Limits
Example session with this skill installed
How much of my Claude limit is left? I want to start a big refactor with a few subagents.
- Read your context and instructions
- Compiled the claude usage limits
Ran python scripts/check_usage.py:
5-hour session: 72% used, resets in 1h 40m
Weekly (all models): 31% used, resets in 4d 12h
Summary
Your 5-hour limit is at 72%, so there is room for the refactor but not for a wide fan-out. The weekly limit is fine.
| Limit | Used | Resets in |
|---|---|---|
| 5-hour session | 72% | 1h 40m |
| Weekly (all models) | 31% | 4d 12h |
Plan
- I will do the refactor with one subagent instead of four, module by module.
- If the 5-hour limit passes 85% midway, I will stop at the next clean point and tell you, so nothing is left half-done.
- The 5-hour window resets in 1h 40m. The full parallel test run can wait until then.
Connects securely to your tools. The creator never sees your data.
About this skill
Ask your agent how much of your Claude limit is left and get one line per limit: percent used and time to reset, for the 5-hour session limit and the weekly limit. The agent can also run it on its own before heavy work, like a large refactor, a long batch job or several parallel subagents, and scale the plan down when you are close to the cap.
What it does
- Reads the 5-hour session limit and the weekly limit for the account Claude Code is signed in with.
- Shows percent used and a countdown to each reset.
- Tells the agent how to size work: go ahead below 60%, fewer parallel agents from 60%, ask you first above 85%, stop at 100%.
--jsongives machine-readable output.
How it works
One Python script, standard library only. It reads the login Claude Code already keeps (~/.claude/.credentials.json on Windows and Linux, the Keychain on macOS), makes one HTTPS request to api.anthropic.com/api/oauth/usage and prints the result. The token goes nowhere else and is never printed or saved.
Known limitations
- Needs Claude Code signed in with a Pro or Max account. An API key alone has no subscription limits.
- The endpoint is undocumented, so Anthropic can change it at any time, and using your login token with it may fall outside Anthropic's terms.
- The endpoint rate limits. The skill tells the agent to check once per task, not in a loop.
- Not affiliated with or endorsed by Anthropic.
On Windows and want the numbers on screen all the time? The free open-source Claude Usage Widget shows the same two limits as a desktop card.
Needs a Claude Pro or Max login in Claude Code; an API key alone has no subscription limits. The endpoint is undocumented and can change without notice.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Free forever
No account required to browse
Trust & safety
Security scan
Security scan passed 5 days ago
- Free with an Agensi account
Requires
Permissions
Allowed Hosts
Reads the Claude Code login (credentials file, or the macOS Keychain) read-only, runs one Python script, and makes one HTTPS request to api.anthropic.com. CLAUDE_CONFIG_DIR is read only to find a non-default config folder. Nothing is written.