> ## Documentation Index
> Fetch the complete documentation index at: https://docs.seekout.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Prompting tips

> Write requests that produce focused searches and are easy to refine.

You do not need special syntax to use SeekOut MCP. Describe the recruiting outcome you need in plain language, then use follow-up requests to refine the result. Apply the following tips whenever they are relevant; they are not sequential steps.

## Be specific

Include the role, location, seniority, and must-have skills. Separate required criteria from preferences so the assistant knows what it can relax.

<Prompt description="Find senior data engineers in Austin with Snowflake experience. Require production data-pipeline experience; fintech experience is preferred." icon="quote">
  Find senior data engineers in Austin with Snowflake experience. Require production data-pipeline experience; fintech experience is preferred.
</Prompt>

## Name the data source when it matters

Let the assistant choose a source for general searches. Name a source when you need a specific kind of evidence or want to compare talent pools.

<Prompt description="Search the GitHub talent pool for senior Rust engineers and cite recent open-source contribution evidence." icon="quote">
  Search the GitHub talent pool for senior Rust engineers and cite recent open-source contribution evidence.
</Prompt>

See [Talent data sources](/mcp/capabilities/data-sources) for the available pools and their searchable signals.

## Refine the current result

Continue in the same conversation when you want to broaden, narrow, or recalibrate a search. State what was wrong with the previous result and what should remain unchanged.

<Prompt description="The results are too junior. Keep the same location and skills, but require experience leading production projects." icon="quote">
  The results are too junior. Keep the same location and skills, but require experience leading production projects.
</Prompt>

Starting a new conversation may lose the search context established by earlier messages.

## Correct the search strategy

If a result or count looks wrong, ask the assistant to explain and adjust its approach instead of only requesting more results.

<Prompt description="The count looks low. Explain which requirement is narrowing the pool, then try a broader title and free-text skill search." icon="quote">
  The count looks low. Explain which requirement is narrowing the pool, then try a broader title and free-text skill search.
</Prompt>

## Ask how the search was built

Request the interpreted criteria, selected data source, and applied filters when you need to audit or reproduce a search.

<Prompt description="Show me the data source, query, and filters used for this result." icon="quote">
  Show me the data source, query, and filters used for this result.
</Prompt>

## Verify must-haves after each refinement

When you refine a search several times, a required criterion can occasionally be dropped along the way. After a refinement, restate your must-have requirements and ask the assistant to confirm each one is still applied — including any power filters, such as security clearance or work authorization.

<Prompt description="Confirm the current search still requires people-management experience and the US work authorization power filter. List every filter that is applied." icon="quote">
  Confirm the current search still requires people-management experience and the US work authorization power filter. List every filter that is applied.
</Prompt>

## Combine related tasks carefully

You can combine search, evaluation, and candidate actions in one request. For actions that use credits or write data, ask the assistant to preview the result before completing the action.

<Prompt description="Find five matching candidates, explain the ranking, and show me which candidates would be saved before adding them to a Recruit workspace." icon="quote">
  Find five matching candidates, explain the ranking, and show me which candidates would be saved before adding them to a Recruit workspace.
</Prompt>

See [Work with candidates](/mcp/capabilities/candidate-actions) for credit usage, write effects, and action requirements.