Conversation Memory No native equivalent

search_all_conversations

Same as search_conversations but searches across all indexed projects simultaneously. Cross-project institutional knowledge.

What It Does

search_all_conversations extends conversation search beyond the current project. Where search_conversations searches the history of one project, this tool searches every indexed project simultaneously.

Why Cross-Project Matters

Solutions found in one project often apply to another. You solved a complex Prisma migration issue while working on Project A. Three weeks later, you hit a similar migration problem in Project B. Without cross-project search, that solution is effectively lost — you would need to remember which project you were working on and manually switch context to search there.

search_all_conversations removes that barrier. A single query searches across all projects:

search_all_conversations("Prisma migration rollback strategy")

Results include the project name so you know where the conversation happened, and the full turn-pair so you can extract the solution.

When to Use

  • Pattern reuse. You are implementing a feature similar to one you built in another project. Search for the design discussion to avoid repeating mistakes.
  • Error diagnosis. An error message or stack trace that you have seen before but cannot remember where. Cross-project search finds it regardless of which project contained the fix.
  • Consistency checks. Before making an architectural decision, search whether the same question was debated in another project and what conclusion was reached.
  • Knowledge transfer. When onboarding to a new project within the same organization, search for conversations about shared libraries, common patterns, or organization-wide conventions.

Scoring

Results from different projects are scored using preserved BM25 magnitudes rather than normalized scores. This means a highly relevant result from Project A will rank above a marginal result from Project B, even though they come from separate indexes. The scoring does not artificially equalize results across projects.

Key Parameters

  • query (required) — natural language search query
  • limit — max results across all projects. Default 10.

Relationship to search_conversations

Use search_conversations when you know the answer is in the current project or a specific project. Use search_all_conversations when you are unsure which project contains the relevant conversation, or when you want to cast a wide net.

Benchmark note

This benchmark compares CodeSift against the closest practical native workflow an agent would use for the same task. For some tools, that baseline is a direct shell equivalent such as rg or find. For AST-aware, graph-aware, and LSP-backed tools, the baseline is a multi-step workflow rather than a strictly identical command. Results should be read as agent-workflow comparisons: token cost, call count, and practical context efficiency.