Codemod grep: Structural code search across public GitHub repos
Codemod grep is a new code search engine that works on structure rather than plain text. It uses ast-grep pattern syntax to match code by its syntax tree shape, so you can find patterns across public GitHub repositories that regex or text search would miss.
What makes it different
Instead of searching for exact strings or fuzzy regex matches, you write patterns that represent code structure. For example, the pattern console.log($MSG) will match any console.log call with any argument. Similarly, $A.$B($C) matches any method call on any object with a single argument, and await $EXPR finds any awaited expression.
The tool also supports structural search for HTML patterns, like finding <img> tags without an alt attribute — something that's tricky with regex but trivial with structural matching.
How to use it
The interface is simple: you enter an ast-grep pattern in the search box, optionally switch to YAML rule mode for more complex conditions, and hit search. There's also a regex mode for traditional string matching, and an "Ask AI" feature for natural language queries.
Under the hood, it leverages the exact same pattern syntax used by ast-grep's CLI, so if you already know how to write ast-grep rules, you can reuse those skills here. For a quick start, the homepage offers example patterns like console.log($MSG), $A.$B($C), and await $EXPR.
Who this is for
This will appeal to developers who work on large codebases, perform cross-repository refactoring, or need to audit code patterns across the open-source ecosystem. It's also a great way to discover how certain APIs are used in practice.
Head over to grep.codemod.com to try it out with your own patterns.
📖 Read the full source: HN AI Agents
👀 See Also

MemAware Benchmark Tests AI Memory Beyond Keyword Search
MemAware is a benchmark with 900 questions across 3 difficulty levels that tests whether AI assistants with memory can surface relevant context when queries don't hint at it. Results show BM25 search scored 2.8% vs 0.8% with no memory, while vector search drops to 0.7% on cross-domain connections.

ForgeAI: A Visual Workbench for Model Engineering
ForgeAI provides a visual interface for model inspection, merging, and training, offering features like 3D model architecture inspection and M-DNA Forge for visual layer merging.

OpenClaw Janitor Skill for Automated System Management and Security Hardening
A developer created a skill that uses Claude Code to SSH into OpenClaw machines and harden configurations, including sandboxing, OS hygiene, and channel security, while maintaining a project folder with audit instructions in CLAUDE.md.

Stanford Researchers Release OpenJarvis: A Local-First Framework for On-Device AI Agents
Stanford researchers have released OpenJarvis, a local-first framework for building on-device personal AI agents with tools, memory, and learning capabilities. The project includes GitHub repository and website links for developers to explore.