How it works
- Connect the Unleash MCP server for your instance at https://<your-instance-url>/api/admin/mcp and sign in through the browser flow, or run the local server with your instance URL and personal access token where your setup calls for that route.
- Install the ExpLint MCP at https://explint.com/mcp to enable your agent to audit tests.
- Run the prompt below to audit these plans before the work starts.
How the check runs
- It reads the flag's description, activation strategies with their rollout share and constraints or segments, and strategy variants with their weights, and it asks you to confirm variant and constraint detail where your server version returns only a summary.
- It sends that text to ExpLint with
explint__lint, which returns errors, warnings, and passes, each finding naming the rule behind it. - You edit the flag in Unleash and ask for a fresh read, keeping the findings in chat unless you ask the assistant to change flag configuration through the MCP tools.
Copy and paste this prompt
Change the words in square brackets to name your own project, board, or test.
Prompt for your assistant
Use the Unleash MCP tools to read the experiment flag [flag name] in project [project name]. Build one text from its description, rollout share, constraints or segments, and variants with their weights, and state which of hypothesis, primary metric, guardrails, run length, and sample size the flag did not carry. Lint it with explint.com against project [your ExpLint project name]. Show each error and warning with its Unleash field. Do not launch or edit the flag yet. Suggest fixes here and re-check after I update Unleash.
What ExpLint looks for
ExpLint ships over 24 rules, and a project switches each one on or off or rewrites it in plain English. These are the rules the plans this guide reads break most often. An error means the test cannot answer its question as written, and a warning means the result will be hard to judge when it comes in.
| Rule | Level | In plain words |
|---|---|---|
| randomization/states-the-unit | error | Unleash splits by whatever its strategy names, and the plan should say whether that unit is a user, a session, or an account. |
| randomization/states-the-split | error | The rollout share and variant weights belong in the plan text, since the split decides whether the test can finish in the time it holds. |
| audience/has-exclusion-criteria | warning | Constraints and segments choose who enters the experiment, and the plan should say separately which accounts it keeps out and why. |
| rollout/has-a-rollback-plan | error | The flag is the switch, and the plan should name who turns it off and what result would make them do it. |
