Introducing Agentic Search for Code and Context

Entire has introduced Agentic Search, an API and CLI-integrated tool that lets coding agents search across accessible repositories for both exact code and the historical reasoning, commits, and session transcripts behind changes. Its dual search engine combines semantic retrieval of discussions and commit context with symbol-aware code search, using turbopuffer for hybrid BM25 and semantic queries with regionally distributed storage, while code indexes are incrementally maintained and cached for roughly 100-millisecond median query latency. In internal tests involving nine engineering-history questions and 90 runs per scenario, agents using search answered 81 of 90 questions correctly versus 70 without it, while reducing average token use from 547,000 to 262,000, costs from $0.38 to $0.23, and average steps from 14 to 7; however, the company notes that the benchmark may favor its own detailed, agent-written commit history. Users can enable the feature through the Entire CLI, search sessions, commits, transcripts, and code across selected repositories, use browser or JSON interfaces, and access definitions and references, as Entire expands its platform as a record of agentic coding work.

No tracked trend matches for this post yet.

Use this post, company, and trend context to find content marketing opportunities, perform competitive analysis, or address product feature gaps via the Plushcap MCP server or the Plushcap API.