Skip to content

Release Notes: v0.1.7

Date: 2026-05-29

This release expands the core lint-ai model beyond document indexing and retrieval into a more general semantic graph for code-oriented workflows. It adds shared symbol, ownership, usage, and review packet structures that can be consumed by language adapters and higher-level review tools. The existing SourceDocument schema remains unchanged in this release.

Highlights

  • Bumped crate version from 0.1.6 to 0.1.7.
  • Added a generic SymbolRecord / SymbolStore layer for symbol indexing.
  • Added a generic OwnershipRecord model for allocation, release, transfer, and borrow facts.
  • Added a generic UsageGraph layer for connecting documents, symbols, and ownership edges.
  • Added a CorpusGraph wrapper so document, symbol, and usage state can be queried together.
  • Added generic review packet and review finding models for AI-assisted code review workflows.

Semantic Graph Layers

The main architectural change in this release is the addition of shared semantic layers alongside the existing document model:

  • SourceDocument
  • the existing raw artifact record
  • unchanged in this release
  • still suitable for Markdown, code, and other source types
  • SymbolRecord
  • language-agnostic symbol metadata
  • supports declarations, references, uses, calls, imports, and inheritance
  • OwnershipRecord
  • language-agnostic lifetime and ownership facts
  • supports allocations, releases, transfers, borrows, aliases, and escapes
  • UsageGraph
  • stores nodes and edges for symbol and ownership relationships
  • supports symbol-centric and document-centric queries
  • CorpusGraph
  • bundles document, symbol, and usage state into one queryable view

Review Model

This release also adds the shared data model needed for AI review packets.

Review packet data

  • file-centric summaries
  • symbol summaries
  • ownership summaries
  • usage summaries
  • review findings and recommendations

Review intent

  • represent changed files as a compact review packet
  • keep the packet structured but queryable
  • let higher-level tooling decide how to surface and rank findings

Why This Matters

lint-ai is now more than a Markdown/document index. It provides the shared semantic layer needed for multi-language symbol search, ownership reasoning, usage tracing, and review-oriented packet generation, while keeping the SourceDocument schema stable.