Prompts

Prompts are pre-built multi-step workflows that guide AI agents through complex tasks. The Enhance MCP server provides two prompts.

investigate

A multi-step codebase investigation workflow. Searches code, knowledge base, and issues to understand how something works.

Parameters

Parameter	Type	Required	Description
query	string	Yes	The question to investigate
projectId	string	Yes	Enhance project ID
repositoryId	string	No	Repository ID (discovered automatically if omitted)

Workflow steps

1. Search the knowledge base for relevant documentation
2. Search the codebase for relevant source files
3. Read key files identified in the search
4. Check the codebase map to understand architecture
5. Search for related issues or past work
6. Synthesize findings into a structured answer

Output

The agent produces a structured response with:

• Summary of how it works
• Key files involved
• Relevant issues and documentation
• Gaps or concerns identified

Example

query: "How does JWT token refresh work?"
projectId: "proj_abc123"

create-issue-from-context

Creates a well-structured issue from code context. Searches the codebase first to understand scope and affected files.

Parameters

Parameter	Type	Required	Description
title	string	Yes	Issue title
description	string	Yes	Context for the issue
projectId	string	Yes	Enhance project ID
repositoryId	string	No	Repository ID (discovered automatically if omitted)

Workflow steps

1. Search for existing similar issues (avoid duplicates)
2. Search the codebase for affected files
3. Analyze scope using the codebase map
4. Create the issue with full context including:
   • Clear, actionable title
   • Detailed description with what needs to change and why
   • Affected files identified from search
   • Appropriate priority, type, and tags
5. Report the created issue and any related issues found

Example

title: "Add rate limiting to public API endpoints"
description: "Public API endpoints currently have no rate limiting, which exposes us to abuse"
projectId: "proj_abc123"