researcher

Overview

The researcher agent investigates external documentation and references. Call it when you need information that isn't in the codebase — things like SDK usage, library API specs, or service constraints. Its role is to secure accurate information before implementation, reducing the chance that executor writes code based on guesses.

When to use

• When you need to verify the correct usage of an SDK or library you haven't used before
• When you need to understand the constraints, rate limits, or authentication method of an external service API
• When official references need to be checked because current docs may differ from training data
• When parallel documentation research is needed via $external-context or $document-specialist

Examples

"Look up how to verify Stripe Webhooks based on the official docs"
"Find how to configure middleware in Next.js App Router"
"Research how to migrate this library from v3 to v4"

Research scope

Process

1. Clarify the research target — what exactly needs to be known.
2. Check official documentation or references first.
3. Use documentation tools like context7 when necessary.
4. Summarize findings in a form that can be used directly for implementation.

Inputs

• Name of the library, SDK, or service to research
• The specific API or feature to verify
• Areas where implementation is blocked or uncertain

Outputs

• Accurate usage summary based on official documentation
• Function signatures, configuration options, and example code
• Known constraints or caveats
• Reference source links

Limits

• Does not write code directly. Provides information and hands off to executor for implementation.
• States uncertainty explicitly when official documentation is unavailable.
• Internal codebase exploration is handled by explore.

Related agents

• executor — implements based on the information researcher retrieves.
• dependency-expert — handles SDK/package selection and version judgment.
• explore — handles internal codebase exploration and file structure mapping.