Release Notes & Changelog
Relevant source files
The following files were used as context for generating this wiki page:
Arbor utilizes a structured changelog and semantic versioning to document the evolution of its Code Property Graph (CPG) engine. The release notes track the transition from an initial AST parser to a production-grade impact analysis system, highlighting advancements in multi-language support, persistence layers, and AI-native integration via the Model Context Protocol (MCP).
Sources: CHANGELOG.md:43-100, PR_DESCRIPTION.md:3-8
Current Version: v1.5.0
The v1.5.0 release focuses on refining the user experience and improving the precision of AI-driven analysis. Key updates include more accurate token counting using tiktoken for LLM context budgets and the introduction of fuzzy symbol suggestions for CLI interactions.
Key Enhancements
- Precision Context: Replaced heuristic token estimation with
tiktoken(cl100k_base) to ensure accurate context slicing for LLMs. - Fuzzy Search: Implemented Jaro-Winkler similarity matching in the CLI to suggest symbols when exact matches fail (e.g., suggesting
authfor the typoautth). - UX Improvements: Added a version watermark to the GUI and improved Python UX by handling empty
__init__.pyfiles silently.
Sources: README.md:12-25, crates/arbor-cli/src/commands.rs:645-660
Version History
[1.1.0] - "The Sentinel Update"
This update introduced advanced impact analysis and refactoring tools designed to predict breakage before code changes are applied.
- Impact Radius Simulator: Implemented bidirectional BFS in
impact.rsto classify affected nodes by severity (Direct, Transitive, Distant). - Dynamic Context Slicing: Added
slice.rsfor token-bounded context extraction, allowing developers to pin critical nodes for LLM prompts. - CLI Additions: Introduced
arbor refactorfor previewing blast radius andarbor explainfor graph-backed code explanations.
Sources: CHANGELOG.md:5-29, crates/arbor-cli/src/commands.rs:488-510
[1.0.0] - "The Logic Forest"
This milestone transformed Arbor into a production-grade engine by introducing persistence and cross-file symbol resolution.
- World Edges: Implemented FQN-based cross-file reference resolution.
- Persistence: Integrated
sleddatabase for local graph storage, utilizingbincodefor binary serialization. - Scale: Added Barnes-Hut QuadTree for force simulation and viewport culling to support codebases with 100k+ nodes.
Sources: PR_DESCRIPTION.md:16-30, CHANGELOG.md:34-53
Feature Evolution Table
The following table summarizes the rollout of core features across major versions:
| Feature | Introduced | Description |
|---|---|---|
| ArborQL (MCP) | v1.0.0 | Shortest path tracing (find_path) for AI agents. |
| Persistence | v1.0.0 | Incremental sync using sled and binary serialization. |
| Impact Radius | v1.1.0 | Bidirectional BFS to predict affected nodes. |
| Context Slicing | v1.1.0 | Token-bounded extraction for LLM prompts. |
| Native GUI | v1.4.0 | Interactive impact analysis interface. |
| Fuzzy Matching | v1.5.0 | Typo-tolerant symbol suggestions (Jaro-Winkler). |
Sources: CHANGELOG.md:34-53, PR_DESCRIPTION.md:10-35, README.md:12-25
Language Support Roadmap
Arbor has rapidly expanded its language coverage, moving from 4 supported languages in the initial release to 10 by version 1.0.0.
Note: This diagram shows the chronological addition of language parsers into the core engine.
Sources: CHANGELOG.md:58-100, PR_DESCRIPTION.md:32-35
AI Integration (MCP)
Arbor's integration with AI agents is a primary design goal, implemented via the Model Context Protocol (MCP). The capabilities provided to AI agents have evolved from basic graph queries to complex impact analysis.
MCP Tool Schema
| Tool Name | Version Added | Description |
|---|---|---|
get_logic_path | v1.0.0 | Traces call graphs for dependencies. |
analyze_impact | v1.1.0 | Computes blast radius of node changes. |
find_path | v1.0.0 | Finds shortest path between two nodes using A*. |
Note: The flow describes how AI agents interact with the Arbor graph through the MCP bridge.
Sources: crates/arbor-mcp/src/lib.rs:163-230, PR_DESCRIPTION.md:11-14
Summary of Changes
The project documentation follows the Keep a Changelog format and adheres to Semantic Versioning. Since its initial release (v0.1.0), Arbor has transitioned from a simple AST visualizer to an "execution-aware" analysis engine capable of providing deterministic, explainable answers regarding code dependencies and refactoring risks.
Sources: CHANGELOG.md:31-32, README.md:65-75