How it works
- Connect the Harness MCP server at https://mcp.harness.io/mcp to your AI assistant or coding agent and sign in with OAuth through Harness ID, or run the current harness-mcp-v2 server locally with a Harness API key where your account 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 experiment's hypothesis, baseline and comparison treatments, key and supporting metrics, start and end times, then takes the traffic allocation and targeting rules from the flag definition that holds them.
- 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 experiment in Harness and ask for a fresh read, keeping any agreed changes inside the organisation and project the experiment belongs to.
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 Harness MCP tools to read the experiment [experiment name] in project [project name]. Build one text from its hypothesis, baseline and comparison treatments, traffic allocation and targeting rule from the flag, key and supporting metrics, and start and end dates. Lint it with explint.com against project [your ExpLint project name]. Show each error and warning with its Harness field. Do not launch or edit the experiment yet. Suggest fixes here and re-check after I update Harness.
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 |
|---|---|---|
| hypothesis/explains-the-mechanism | warning | The hypothesis should say why the change should move people, since Harness records the treatments without recording the reasoning behind them. |
| metrics/has-one-primary-metric | error | The key metric decides the outcome, and the supporting metrics act as guardrails only when the plan says which of them has a limit. |
| randomization/states-the-split | error | Allocation lives in the flag, and the plan still has to state each treatment's share for the reader judging the run length. |
| duration/covers-full-business-cycles | error | The start and end times should span full weeks, because a three-week window weighs weekdays and weekends differently from a full cycle. |
