EgentCodePlexus

•
10
•

Provides a high-performance code intelligence graph for LLMs and AI agents, enabling sub-second structural queries, impact analysis, and cross-repo API contracts.

EgentCodePlexus serves as a critical structural-knowledge layer for autonomous AI coding agents, designed to be invoked dozens of times per task. Unlike traditional IDE-focused tools, it delivers token-cheap, non-interactive outputs tailored for agent consumption, with zero warm-up and no persistent daemon. It offers unparalleled performance, indexing large polyglot repositories in seconds and returning complex queries in milliseconds, while explicitly flagging unresolved dependencies to prevent AI hallucinations. This polyglot tool supports 31 languages, providing deep structural insights necessary for modern multi-stack development.

主な機能

01Sub-second query and rapid cold indexing performance for large, polyglot repositories.
02Emits `BlindSpot` records for honest unknown dependencies, preventing AI hallucinations.
03Agent-centric design with token-cheap, non-interactive outputs and no persistent daemon.
04AST-aware multi-file rename, impact analysis, and cross-repo API contract inventory.
05Polyglot support for 31 programming languages, including deep structural coverage for 14.
063 GitHub stars

ユースケース

01Autonomous code refactoring and symbol renaming across multiple files and languages.
02Performing rapid blast-radius analysis to understand the impact of code changes.
03Inventorying and analyzing HTTP routes and API contracts across multi-stack repositories.