Weld FAQs

Question 1

How does Weld benefit AI agents and LLMs?

Accepted Answer

Weld is agent-native, providing an MCP server that allows AI agents (like Claude Code and Codex) to directly query the codebase's knowledge graph. This enables agents to efficiently understand context, find relevant information, and perform complex tasks.

Question 2

What kind of information can Weld discover within a codebase?

Accepted Answer

Weld performs whole-codebase discovery, going beyond just source code. It extracts context from documentation, configuration files, CI workflows, infrastructure definitions, and build files, providing a comprehensive view.

Question 3

What is Weld and its primary purpose?

Accepted Answer

Weld is an agent-first tool designed to link context, actions, and knowledge within a codebase. It forms a connected structure to enable intelligent repository discovery and navigation for both developers and AI agents.

Question 4

Is Weld compatible with multiple programming languages and customizable?

Accepted Answer

Yes, Weld supports multiple languages with bundled Tree-sitter strategies for popular options like Python, Go, Rust, and C#. It's also highly customizable via a config-driven `.weld/discover.yaml` and a plugin architecture for custom extraction strategies.

Question 5

What is Weld's 'Polyrepo Federation' feature?

Accepted Answer

Polyrepo Federation allows Weld to manage and connect knowledge graphs across multiple nested Git repositories within a single workspace. It builds a meta-graph of cross-repo relationships, making it ideal for complex polyrepo architectures.

Weld

Weld

主要功能

使用案例

主要功能

使用案例