Skip to content

ROADMAP.md

Latest commit

 

History

History
74 lines (56 loc) · 4.72 KB

ROADMAP.md

File metadata and controls

74 lines (56 loc) · 4.72 KB
Raw

Arbor Roadmap: Path to v2.0 & Beyond

Vision: Arbor is the "Nervous System" for AI Agents—a persistent, visual, and intelligent memory graph that prevents hallucinations and enables safe, massive-scale refactoring.

✅ Current Execution Status (2026-05-30)

This section tracks what is already shipped versus what remains strategic work.

Shipped / In-Progress Foundation (as of 2026-05-30)

  • PR Intelligence & Automation (v2.2.0): Native markdown output formatting for impact reports (arbor diff/check --markdown), automatic PR auto-descriptions (arbor summary), upgraded composite GitHub Action with automatic comment deduplication/in-place editing, and GitHub-native sponsorships configuration (.github/FUNDING.yml).
  • v2.1.0 Agent-Native MCP: 6 new surgical tools (list_entry_points, get_callers, get_callees, search_symbols, get_file_graph, get_node_detail) with standard JSON envelope and suggested_next_tool routing hints. Total: 10 MCP tools.
  • Agent Bridge (MCP): Operational with expanded tools (get_knowledge_path with logic paths/Markdown links + explanations, analyze_impact with markdown table format for PR bot, find_path, sorted_by_centrality + ConfidenceExplanation)
  • Tauri Pivot/Lattice: Desktop shell with system tray (Personal OS feel), integrated graph/MCP (companion starter startup in Arbor v2/)
  • arbor audit + impact workflows: Git-aware, with blast radius for code PRs
  • Language expansion: Full with fallback_parser for Markdown (NodeKind::Section), Dart fixes, all tests passing
  • Persistent store: Sled in GraphStore with incremental updates, centrality precompute
  • VS Code extension, releases, distribution: Updated with PR bot foundation

Still Outstanding (Major Epics)

  • Full write-mode visualizer + 4D time-travel (Priority for 4 features)
  • Plugin/WASM system for parsers
  • Enterprise RBAC, compliance reports, learning loop
  • Deeper monorepo integration for Tauri + arbor crates

🧠 1. Architectural Memory Graph (Visual Intelligence)

Turn impact analysis into a persistent, explorable map.

  • Persistent Graph Store: Move beyond ephemeral indexing to a persistent database (SQLite/Sled) for instant load times.
  • Visual Dependency Explorer: Interactive, queryable UI to answer "What breaks if I delete this?"
  • Time-Travel Analysis: Track architectural drift over time (integration with Git history).

🤖 2. AI Explanation Layer

Make the graph human-readable and trustable.

  • Narrative Engine: Convert raw graph data into sentences (e.g., "This function affects 6 downstream services...").
  • Confidence Contracts: SLA for analysis certainty (e.g., "100% static certainty" vs "80% heuristic").
  • Agent Bridge (MCP): Deepen integration with Claude/Cursor to act as the "ground truth" for AI coding agents.

🛡️ 3. Security & Audit ("Blast Radius for CVEs")

Penetrate the security market with vulnerability tracing.

  • arbor audit <function>: Trace tainted inputs and vulnerable execution paths.
  • Compliance Reports: Generate artifacts for SOC2/ISO 27001 showing impact analysis.

🌍 4. Multi-Language & Ecosystem

Be the #1 tool for every stack.

  • Language Expansion: Full support for JS/TS, Python, Go, Rust, Java, C#.
  • Language Expansion (v1.6.2): Added pragmatic parser support for Kotlin, Swift, Ruby, PHP, and Shell.
  • Plugin System: Wasm-based plugin architecture for community parsers.
  • "Bounty Board": Gamified community contributions for new language parsers.

🏢 5. Enterprise Mode

Features for global dominance.

  • Air-Gapped Support: Fully offline operation (already core, but explicit support).
  • On-Premise Deployment: Dockerized containers for enterprise CI/CD.
  • Role-Based Access: Graph views tailored for Junior vs Senior devs vs Architects.

🔄 6. Continuous Learning Engine

From rule-based to intelligent.

  • Feedback Loop: Learn from user corrections ("No, this isn't a dependency") to improve heuristics.
  • Pattern Recognition: Automatically detect and adapt to repo-specific architectural patterns (e.g., "All *Service classes are singletons").

🚀 Immediate Focus (v1.8 / v2.0)

Theme: Persistent Intelligence & Enterprise Readiness

  1. Persistent Graph Database: Move to SQLite/Sled with transactional updates and indexed queries for instant load.
  2. Plugin System (Wasm): Community-extensible parser and runtime plugins.
  3. Time-Travel Analysis: Git-history-aware architectural drift detection.
  4. Security Layer Expansion: Deepen arbor audit with taint-style path confidence and compliance report generation.