Back to Signal Feed
PaperTracked since Apr 22, 2026

Sourcegraph clarifies when to use Code Search, Deep Search, and MCP

Sourcegraph published guidance that distinguishes Code Search, Deep Search, and MCP as separate tools and maps each to the kind of code-finding task it should handle, helping teams avoid treating the three as interchangeable.

SourcegraphCode SearchDeep SearchMCP

What Happened

  • Sourcegraph published guidance that distinguishes Code Search, Deep Search, and MCP as separate tools and maps each to the kind of code-finding task it should handle, helping teams avoid treating the three as interchangeable.
  • Sourcegraph published guidance that distinguishes Code Search, Deep Search, and MCP as separate tools and maps each to the kind of code-finding task it should handle, helping teams avoid treating the three as interchangeable.
  • 1 evidence item attached for review.

What is Different

Before

Scattered source updates, isolated context, and manual follow-up across multiple feeds.

Now

Introduced a clear decision framework for choosing among three Sourcegraph search options by query intent and workflow type, which reduces confusion about when to use lexical search, semantic search, or MCP-assisted retrieval.

Why Track This

Why It Matters

Developers and operators can reduce wasted time by selecting the right search mode for each task, so code discovery and debugging become faster and less frustrating on large codebases. The practical change is a more predictable workflow: quick exact matches via Code Search, broader intent-based exploration via Deep Search, and MCP where model-driven context retrieval is needed. Next, monitor whether teams consistently apply the guidance, whether Deep Search stays accurate and fast enough at scale, and whether MCP rollout is limited by integration or access constraints in enterprise environments.

Impact

Developers and operators can reduce wasted time by selecting the right search mode for each task, so code discovery and debugging become faster and less frustrating on large codebases. The practical change is a more predictable workflow: quick exact matches via Code Search, broader intent-based exploration via Deep Search, and MCP where model-driven context retrieval is needed. Next, monitor whether teams consistently apply the guidance, whether Deep Search stays accurate and fast enough at scale, and whether MCP rollout is limited by integration or access constraints in enterprise environments.

What To Watch Next

  • Watch whether Sourcegraph becomes a repeated pattern.
  • Track follow-up changes around Enterprise Search.
  • Compare future signals against this evidence trail.
  • Re-check risk flags: teams_misclassify_tasks_and_choose_wrong_search_mode, deep_search_performance_regresses_on_large_repos.
Open Topic TimelineOpen Technical EventOpen Original Sourceteams_misclassify_tasks_and_choose_wrong_search_mode / deep_search_performance_regresses_on_large_repos / mcp_requires_additional_configuration_and_permissions

Supporting Evidence

RSS FEEDHigh Trust

Code Search, Deep Search, or MCP: When to Use Each

AI added new ways to search code, but not all of them apply to every problem.