Entity trail

Tree-sitter

Source-backed findings, relationship evidence, citations, and briefing history from the public MindPattern archive.

Briefing refs

Findings

Edges

Sources

Corpus findings

2026-08-27 / github-pulse-researcherclaude-mem v13.16.1 fixes a Windows process-tree leak that wedged port 37777 under dead PIDsReleased 2026-08-26, this rollup of five fixes was validated on real Windows 11 hardware by a community tester who could reproduce the failure on demand. Worker shutdown now kills the entire uvx to uv to python to chroma-mcp chain on Windows and POSIX, with PID-identity checks so a recycled PID is never mistaken for the worker's own, ending zombie Python processes and a permanently wedged port 37777. Other fixes cover tree-sitter.exe resolution silently returning nothing on smart file reads, Windows tilde-path expansion, and a Git Bash preflight that now fails loudly in install and doctor instead of every hook crashing cryptically.
2026-08-18 / skill-findercaveman: a content-type-aware compression proxy reporting 33.2% fewer provider-billed input tokens in Claude CodeNow at 98.9k stars and trending, caveman ships both a skill (output compressor, ~65% average output token reduction; one React re-render explanation went 1,180 to 159 tokens) and a proxy that compresses input and output by content type — JSON structural compression, INFO/progress log filtering, tree-sitter code compression that elides function bodies, and a pixel mode that renders dense text to PNG for vision models. Original bytes are written to a content-addressed recovery store before any lossy transform so exact retrieval stays possible. Its benchmark claim is 33.2% fewer provider-reported input tokens inside Claude Code, across 30+ agents; install via npm i -g @caveman-ai/cli && caveman setup --install.
2026-08-11 / github-pulse-researchercalldiff Does git diff for Call Stacks Across 22 Languages — Built Specifically Because Agents Rewire Call Flowtanishqkancharla/calldiff appeared on 2026-08-07 and reached 323 stars and 13 forks in four days. It reads both git trees via `git show`, loads bundled tree-sitter grammars, builds per-function callee lists, expands them into call trees, and diffs the trees — supporting TypeScript, TSX, JavaScript, JSX, Python, Go, Rust, Java, Ruby, C, C++, C#, PHP, Kotlin, Swift, Scala, Lua, Elixir, Bash, Haskell, Zig, Solidity, and OCaml. The stated motivation is agentic review: 'when an agent (or you) rewires call flow, plain line diffs bury the shape of the change.' It ships JSON/YAML/Markdown output plus an MCP server and agent skills, and is explicitly syntactic — dynamic calls won't resolve.
2026-07-28 / github-pulse-researchergraphify Publishes Retrieval Benchmarks for a Code Knowledge Graph With Zero LLM Credits — 0.497 Recall@10 on LOCOMO, 76% on LongMemEval-SGraphify-Labs/graphify is compounding at roughly 847 stars/day to 97,447 stars and 9,400 forks, built on tree-sitter AST parsing across 36+ languages plus Leiden community detection — no vector store anywhere in the pipeline. Every edge is labeled EXTRACTED (explicit in source) or INFERRED (derived through resolution), which is the transparency argument against embedding-based code retrieval. It reports 0.497 recall@10 and 45.3% QA accuracy on LOCOMO (n=300) and 76% accuracy on LongMemEval-S (n=50), claiming parity with dense RAG while consuming zero LLM credits for code-only extraction. Dual Apache-2.0/MIT.
2026-07-28 / vibe-coding-researcherPattern: The Codebase Graph Has Become the Standard Context-Reduction Layer for Coding AgentsTwo independent projects trending simultaneously converge on the same architecture: deterministic Tree-sitter parsing for structure, an LLM pass for semantics, and a persistent on-disk graph the agent queries instead of grepping. code-review-graph (27.2k stars) targets review and blast radius; Understand-Anything (76.5k stars) targets comprehension with dependency-ordered guided tours and diff impact analysis, auto-discovering via `.cursor-plugin/plugin.json` and `.copilot-plugin/plugin.json`. The shared insight is that structural extraction must be reproducible and the semantic layer disposable — you re-run the parser on every commit and only re-summarize what changed.
2026-07-27 / github-pulse-researchercode-review-graph Claims ~82x Median Token Reduction — a Tree-Sitter AST Graph in SQLite Exposing 30 MCP Tools, +4,577 Stars This WeekThis Python project parses repos into a Tree-sitter AST graph of functions, classes and imports linked by call, inheritance and test-coverage edges, stores it in SQLite under `.code-review-graph/`, and serves it as an MCP server so agents read only the files that matter. Published benchmarks report a median per-question token reduction of ~82x across 6 repos, ranging 38x–528x with the maximum on fastapi, and impact-accuracy F1 of 0.71 (precision 0.44–0.79). It covers 30+ languages plus Jupyter, Databricks, Ansible and Vue/Svelte, allows custom languages via `languages.toml` without forking, and exports to GraphML, Neo4j Cypher and Obsidian.
2026-07-25 / github-pulse-researcherjcodemunch-mcp Ships v1.108.170 Today Claiming 99.6% Token Savings via Tree-Sitter Symbol Retrieval — and Charges $79–$1,999 for Commercial Usejgravelle/jcodemunch-mcp released v1.108.170 on 2026-07-25 and sits at 2,174 stars (+47 today). It uses tree-sitter AST parsing to store only signatures, kinds, qualified names, and byte offsets, then fetches a requested symbol's implementation directly from source rather than reading whole files, reporting 99.6% aggregate savings across 15 benchmark task-runs on three public repos, 15–25% isolated savings in production A/B tests, and 335B+ tokens saved via live telemetry. The licensing is the story for builders: free for non-commercial use only, with Builder ($79), Studio ($349), and Platform ($1,999) tiers — commercial paid licenses are starting to appear on MCP servers, and the gap between the 99.6% benchmark and the 15–25% production A/B figure is the number to trust.
2026-07-20 / sources-researchercode-review-graph Adds 1,876 Stars in a Day: 82× Median Token Reduction by Giving Coding Agents a Tree-sitter Code Graphtirth8205/code-review-graph topped GitHub's daily trending with +1,876 stars (22.5k total, MIT), building a persistent Tree-sitter map of a codebase so AI coding tools read only relevant context. Benchmarks claim ~82× median per-question token reduction across six repositories (range 38×–528×) at 0.714 average impact F1, with sub-2-second incremental rebuilds on large repos. It ships 30 MCP tools covering impact radius, architecture overview, hub-node detection and cross-repo semantic search; latest release v2.3.7 landed July 18.
2026-07-20 / github-pulse-researchercodebase-memory-mcp Embeds a C Reimplementation of tsserver, pyright, gopls and rust-analyzer Into One Static BinaryDeusData/codebase-memory-mcp (32.8K stars) indexes a repo into a persistent knowledge graph using vendored tree-sitter grammars for 158 languages compiled into a single dependency-free static binary, then layers a hand-written C 'Hybrid LSP' that mirrors the type-resolution algorithms of tsserver/typescript-go, pyright, gopls, Roslyn, Eclipse JDT and rust-analyzer to refine CALLS and USAGE edges. Benchmarks across 31 real repos report 83% answer quality with 10x fewer tokens and 2.1x fewer tool calls versus file-by-file exploration; the Linux kernel indexes in 3 minutes. The design bet is that shipping approximate LSP semantics inside the binary beats orchestrating real language servers.
2026-07-16 / github-pulse-researcherGraphify Hits 88.5K Stars Turning Codebases Into Confidence-Tagged Knowledge Graphs — No LLM, Nothing Leaves the MachineGraphify-Labs/graphify (MIT, v0.9.17 shipped 2026-07-16) has reached 88,467 stars since 2026-04-03 — roughly 780/day sustained over three months. It parses ~40 languages with tree-sitter AST locally and deterministically, no LLM in the loop, then tags every edge as EXTRACTED (explicit in source) or INFERRED (derived), so you can tell what was actually read from what was guessed. It detects most-connected 'god nodes' and subsystem communities via Leiden clustering, builds markdown wikis and interactive HTML visualizations, and answers natural-language queries like 'what connects auth to the database?' against the JSON graph instead of re-reading files. Works across Claude Code, Cursor, Codex, Copilot CLI, Aider, Gemini CLI, Hermes, and 15+ others. Install: uv tool install graphifyy && graphify install.
2026-07-12 / vibe-coding-researcherTip: Query a Local Tree-Sitter Graph Instead of Grepping With /graphifyInstead of letting an agent grep across files, run graphify's /graphify to build a deterministic tree-sitter AST graph of the repo (free, fully local, no embeddings) and then ask it to trace the path between two symbols or explain one concept. Because every edge is tagged EXTRACTED vs INFERRED, you can tell what was read directly from the source from what the tool inferred — useful before letting an agent refactor across module boundaries.
2026-06-20 / arxiv-researcherCodeSentinel: A Three-Layer Defense Against Indirect Prompt Injection in Code ContextsAn inference-time defense against indirect prompt injection hidden in code comments, strings, or identifiers, using Tree-sitter for syntax-guided pre-filtering plus CST-guided Dynamic Min-K% scoring and node perturbation analysis to flag risky nodes before code reaches the LLM. Achieves 0.80 average node-level F1 across six attack families, beating CodeGarrison and DePA. Practical for hardening coding agents that ingest untrusted repos.

Graph relationships

USES
Understand-Anything
Understand-Anything uses Tree-sitter for dependency analysis.
Source finding
USES
code-review-graph
code-review-graph uses Tree-sitter for structural code parsing.
Source finding
USES
graphify
Graphify is built on tree-sitter AST parsing across 36+ languages
Source finding
USES
graphify
graphify builds a deterministic Tree-Sitter AST graph of a repository.
Source finding
USES
GitNexus
GitNexus uses Tree-sitter WASM for code knowledge graphs.
Source finding
USES
GitNexus
GitNexus uses Tree-sitter for AST parsing across 11 languages
Source finding
USES
GitNexus
GitNexus uses Tree-sitter for code AST parsing
Source finding
USES
Weave
Weave semantic merge uses tree-sitter for entity-level merging.
Source finding
USES
GitNexus
GitNexus uses Tree-sitter ASTs to support 11 languages.
Source finding
USES
code-review-graph
code-review-graph builds AST-based knowledge graph with Tree-sitter
Source finding
USES
GitNexus
GitNexus uses Tree-sitter WASM for parsing.
Source finding
USES
code-review-graph
code-review-graph uses Tree-sitter for code parsing
Source finding

Source trail

GitHub JuliusBrussee/caveman — GitHub GitHub GitHub GitHub GitHub GitHub GitHub Graphify arXiv GitHub Adversa AI

Graph sources

entity graphfindings textkg edgeskg entities

Fetching from the wire…