Search

Open Source — GitScrum MCP Server is open source under the MIT license. Available on npm and GitHub. Model Context Protocol server for GitScrum — Claude, GitHub Copilot, Cursor, and any MCP-compatible client full operational access to your project management stack.

The search tool provides 1 action — a unified global search that queries across all GitScrum entities in a single call. Instead of searching tasks, discussions, and projects individually, the search tool casts a wide net across your entire workspace and returns matching results from every entity type.

This is the tool your AI assistant reaches for when your request is exploratory — "find everything about authentication", "what do we have related to the billing module", or "search for anything @sarah worked on". It returns results from tasks, projects, discussions, user stories, and more, ranked by relevance.

Actions Overview

Searching

The search action performs a cross-entity query across your workspace. Pass a search string and optionally narrow results by project, entity type, or result count.

Required Parameters

Optional Parameters

Entity Types

When using the type parameter, specify one of these entity types to narrow results:

Basic Search

The simplest usage — pass a search string and get results across all entity types.

You:  "Search for anything about authentication"
AI:   Calls search action=search with q="authentication"
      → returns matching tasks, discussions, wiki pages, and more

You:  "Find everything related to the billing module"
AI:   Calls search action=search with q="billing"
      → returns cross-entity results ranked by relevance

You:  "Search for 'memory leak'"
AI:   Calls search action=search with q="memory leak"
      → returns tasks, comments, and discussions mentioning memory leaks

Scoped Search

Narrow your search to a specific project or entity type for more targeted results.

By Project

You:  "Search for 'login' in the Backend project"
AI:   Calls search action=search with q="login", project_slug="backend"
      → returns results only from the Backend project

You:  "Find references to 'payment' in the Mobile App project"
AI:   Calls search action=search with q="payment", project_slug="mobile-app"

By Entity Type

You:  "Find tasks related to API performance"
AI:   Calls search action=search with q="API performance", type="task"
      → returns only matching tasks

You:  "Search discussions about the database migration"
AI:   Calls search action=search with q="database migration", type="discussion"
      → returns only matching discussion threads

You:  "Find wiki pages about deployment"
AI:   Calls search action=search with q="deployment", type="wiki"
      → returns only matching wiki pages

You:  "Search for user stories mentioning 'checkout flow'"
AI:   Calls search action=search with q="checkout flow", type="user_story"

Combined Filters

You:  "Search for 'timeout' tasks in the Backend project"
AI:   Calls search action=search with q="timeout", type="task",
      project_slug="backend" → returns precisely scoped results

You:  "Find discussions about 'auth' in the API project, show top 5"
AI:   Calls search action=search with q="auth", type="discussion",
      project_slug="api", per_page=5

Multi-Entity Search Patterns

Global search is most powerful when you're exploring a topic across entity boundaries. Here are patterns that leverage cross-entity search effectively:

Discovery

When you don't know where information lives, search globally first, then drill into specific results.

You:  "What do we have about OAuth2?"
AI:   Calls search with q="OAuth2" → returns:
      - 3 tasks related to OAuth2 implementation
      - 1 discussion about OAuth2 strategy
      - 2 wiki pages documenting OAuth2 flow
      AI summarizes findings across all entity types

You:  "Find everything from last week about the mobile release"
AI:   Calls search with q="mobile release" → aggregates results from
      tasks, discussions, comments, and sprints

Impact Analysis

Before making changes, search for all references to understand the scope of impact.

You:  "Search for all references to the v2 API endpoint"
AI:   Calls search with q="v2 API" → identifies tasks, wiki pages,
      and discussions that reference the deprecated endpoint

You:  "Find everything that mentions the user_profiles table"
AI:   Calls search with q="user_profiles" → returns code tasks,
      wiki documentation, and architecture discussions

Context Gathering

Quickly gather all context about a topic before starting work.

You:  "I'm starting work on the payment refund feature. Find all related items."
AI:   Calls search with q="payment refund" → returns:
      - Existing tasks for the feature
      - Discussion threads with design decisions
      - Wiki pages with payment architecture docs
      - User stories describing the requirement

Team Knowledge

Search for contributions and context from specific team members.

You:  "Search for items mentioning @sarah's auth work"
AI:   Calls search with q="auth" → AI filters results by @sarah's involvement

You:  "What decisions have been made about the caching layer?"
AI:   Calls search with q="caching" → identifies discussions and comments
      where decisions were recorded

Search Results

The search tool returns structured results that include:

• Entity type — What kind of object matched (task, discussion, wiki, etc.)
• Title — The name or title of the matching entity
• Excerpt — A content snippet showing where the match occurred
• Metadata — Project, creation date, status, and other relevant fields
• UUID — Unique identifier for drilling into the result with entity-specific tools

The AI assistant can use these UUIDs to fetch full details with the appropriate tool — task action=get, discussion action=get, wiki action=get, etc.

You:  "Search for 'rate limiting' and show me the details of the first task"
AI:   Calls search → gets results → calls task action=get on the first task UUID
      → returns complete task details

Search vs Entity-Specific Filtering

Both approaches have their place. Use the right tool for the job:

Rule of thumb: If you know you're looking for tasks, use task action=filter. If you're exploring a topic across everything, use search.

Context Auto-Resolution

The search tool supports automatic context resolution. If your AI assistant already knows the workspace from a previous call in the conversation, you can say "search for X" without specifying the workspace every time.

When search results include entities from multiple projects, the response includes project identifiers for each result, allowing the AI assistant to drill into any specific project context.

Next Steps

• Tasks: Use task-specific filtering for granular task search.
• Wiki: Search within wiki pages for project documentation.
• Discussions: Browse and create discussion threads.
• Quick Start: Set up the MCP server if you haven't already.