Tree-sitter
PulseAugur coverage of Tree-sitter — every cluster mentioning Tree-sitter across labs, papers, and developer communities, ranked by signal.
9 day(s) with sentiment data
-
Local code search tool 'ctx' combines BM25 and dense vectors
The article details the inner workings of `ctx`, a TypeScript command-line tool designed for local code search. It explains how `ctx` combines BM25 lexical search with dense vector embeddings, fused using reciprocal ran…
-
Code-review-graph tool optimizes AI code reviews by reducing token usage
The code-review-graph tool aims to reduce token consumption for AI code reviews by creating a local graph of a codebase. This graph, built using Tree-sitter, provides AI assistants like Claude Code, Codex, and Cursor wi…
-
AI coding agents perform better with interface skeletons than full dependency files
A new benchmark study reveals that providing AI coding agents with full dependency files, rather than just interface skeletons, can significantly reduce accuracy. The experiment, conducted within isolated Proxmox VMs, s…
-
Graphify tool creates code knowledge graphs for AI assistants
Graphify has released an open-source tool that allows developers to create a code knowledge graph for their projects, which can then be integrated with AI coding assistants like Cursor and Claude Code. This graph provid…
-
Graphify connects code knowledge graphs to OpenClaw, JetBrains, and Gemini CLI
Graphify has released plugins for OpenClaw, JetBrains AI Assistant, and Gemini CLI, enabling these tools to access a code knowledge graph. The OpenClaw and JetBrains integrations connect to Graphify Cloud via an API, al…
-
New tool token-context-mcp solves coding agent 'blind spot'
A new tool called token-context-mcp has been developed to address the "Function Body Blind Spot" faced by autonomous coding agents. Unlike static repository packing methods that prune code to signatures, this local, clo…
-
OneGraph uses Gemma 4 and dependency graphs for LLM code reviews
A new tool called OneGraph aims to improve LLM-based code reviews by grounding them in a real dependency graph rather than relying solely on the LLM's interpretation. Developed using Graphify, a dependency graph extract…
-
New Tool Lets AI Explore GitHub Repos Without Manual Cloning
A developer has created `github-repo-mcp`, a local server designed to allow AI assistants to explore GitHub repositories without manual cloning. The tool acts as a Model Context Protocol server, enabling AI models to in…
-
Graphify uses ASTs for deterministic code navigation in AI agents
Graphify is a new tool that uses abstract syntax trees (ASTs) to create queryable knowledge graphs for code navigation, offering structural certainty over the approximate proximity of vector embeddings. It parses code u…
-
AI Agents: Code Graphs, ReAct Bottlenecks, and Predictive Forecasters
This cluster of articles explores advancements in AI agent architecture and development. One piece details how deterministic tree-sitter parsing can build queryable code graphs, suggesting graph topology is superior to …
-
AI CLI Agents Optimized with Tree-Sitter Token Pruning in Rust
This article details a method for optimizing AI CLI agent performance by pruning redundant API prompt tokens. It utilizes the Tree-sitter parsing library within a Rust implementation to achieve this optimization. The go…
-
AI agents use deterministic tools for reproducible code analysis
The author explores methods for using AI agents to investigate codebases, emphasizing the importance of reproducible results and managing context overflow. The approach involves leveraging deterministic tools like ripgr…
-
AI coding tools can cause architectural amnesia; ADRs offer a solution · 3 sources tracked
Developers are finding that AI coding tools can lead to "architectural amnesia," where established decisions are forgotten and new suggestions contradict past choices. One solution involves embedding Architecture Decisi…
-
Code RAG chunking shifts from token limits to semantic boundaries
A new approach to chunking code for Retrieval Augmented Generation (RAG) systems aims to improve accuracy by splitting code at semantic boundaries like function and class definitions, rather than arbitrary token counts.…
-
Local RAG tool TraceFind detects documentation lies in codebases
Angellina Joyce Paul developed TraceFind, a local Retrieval-Augmented Generation (RAG) system designed to identify discrepancies between code documentation and actual implementation. The tool indexes codebases and docum…
-
jCodeMunch cuts AI token use by 97% via smarter code retrieval
jCodeMunch, an open-source MCP server, offers significant token savings for AI coding agents by using Tree-sitter to retrieve specific code structures rather than entire files. This approach not only reduces API costs b…
-
New tool 'dekko' cuts AI token usage for code analysis by up to 200x
A developer has created an open-source tool called `dekko` to address inefficiencies in how AI models like Claude process large codebases. The tool generates a static code map, utilizing `tree-sitter` for parsing instea…
-
New framework turns vulnerability history into executable detection rules
Researchers have developed BUGSTONE-E2E, a framework designed to transform historical vulnerability data into executable detection rules. This system mines fixing commits to create reusable rules, which are then process…
-
Model Context Protocol's stateless update breaks local Python tools
The Model Context Protocol (MCP) recently underwent a significant architectural overhaul, transitioning to a stateless HTTP request/response model. This change, while beneficial for large-scale distributed agent infrast…
-
Code knowledge graph tools Graphify, GitNexus, and CodeGraph compared
Three code knowledge graph tools—Graphify, GitNexus, and CodeGraph—are evaluated for their ability to serve AI coding agents by indexing repository symbols and their relationships. The article emphasizes that raw capabi…