v0.2.1

[0.2.1] - 2026-07-24

Added

• Added codetrip mcp setup for one-command integration with Codex, Claude Code, Cursor, VS Code/Copilot, and GitHub Copilot CLI.
• Added automatic client detection, --dry-run, targeted client setup, guarded replacement, and Cursor configuration backup.
• Added exact symbol-name resolution to context, impact, traverse,
  path, and rename, with explicit candidate lists for ambiguous names.
• Added impact --format tree for human-readable reverse dependency trees and
  symbol/file summaries while retaining JSON as the default.

Changed

• Improved README onboarding around coding-agent workflows and MCP setup.

v0.2.0

[0.2.0] - 2026-07-23

Added

• Production semantic analysis for Go, Java, Kotlin, C#, Python, Rust, Swift,
  TypeScript/TSX, JavaScript/JSX, PHP, C, and C++.
• Typed CALLS, EXTENDS, IMPLEMENTS, OVERRIDES,
  METHOD_IMPLEMENTS, and DISPATCHES_TO relationships with cross-file
  resolution.
• Repository-scoped symbol search, source search, semantic vectors, hybrid
  retrieval, bounded graph traversal, and shortest paths.
• code, docs, and all source-search scopes.
• context symbol intelligence with source excerpts and typed direct
  relationships.
• impact reverse semantic dependency analysis with depth, confidence, and
  relationship details.
• check graph integrity, inheritance-cycle, import-cycle, and optional
  low-confidence relationship analysis.
• diff Git changed-line mapping to persisted symbols with aggregated impact
  and change causes.
• Non-mutating rename planning with identifier validation, conflict
  detection, semantic references, and exact textual candidates.
• Deterministic parser-inspection CSV and complete persisted-graph CSV export.
• Public Go Engine APIs for all supported library capabilities.
• Single-word CLI commands and matching stdio MCP tools.

Changed

• Rebuilt ingestion around one production analysis path shared by the Go
  library, CLI, and MCP.
• Isolated graph, symbol, source, and vector data per repository.
• Repository replacement now builds and atomically publishes a complete
  snapshot before retiring the previous snapshot.
• MCP is hosted by the CLI adapter and invokes only public Engine methods.
• Traversal results now include edges, relationship types, directions,
  confidence, and resolution reasons.
• Source search defaults to code and engineering configuration instead of
  unrestricted repository text.
• Updated README architecture diagram and English and Chinese user guides for
  the current engine capabilities.

Fixed

• Improved cross-file calls, interface implementations, inheritance, and
  method override resolution across all supported languages.
• Restored PHP Composer namespace, inheritance, implementation, override, and
  call resolution on large framework repositories.
• Fixed Go interface method sets and cross-file call chains.
• Fixed C# generic inheritance and explicit override handling.
• Fixed Python named imports and default-argument calls, Rust workspace
  imports, Swift package visibility and generic constraints, and C/C++ symbol
  resolution.
• Removed dangling community memberships, self imports, invalid self semantic
  relationships, and relationships with missing endpoints.
• Prevented one repository from locking unrelated repository data.
• Added a portable Windows source-search backend while preserving the public
  search behavior.

Removed

• Removed the obsolete community, ingestion, route, wiki, group, collection,
  and incremental-index implementations replaced by the production engine.
• Removed analysis-mode branching; language tuning now uses the single
  production path and optional local CSV validation.

v0.1.0

[0.1.0] - 2026-06-22

Added

• Core Engine: Hybrid Graph-Augmented Code Intelligence Engine with embedded Pebble storage
• Multi-language Parsing: Tree-sitter based parsing with providers for Go, TypeScript, JavaScript, Python, Java, Rust, C, C++, C#, and Markdown
• Knowledge Graph: 38 node label types with typed edges, adjacency indexes, and scope resolution
• Cypher Query Engine: Declarative graph traversal with Volcano iterator model and timeout protection
• Hybrid Search: BM25 full-text search (Bluge) + HNSW semantic vector search with RRF fusion
• Impact Analysis: Upstream/downstream dependency tracing with risk assessment
• Taint Explanation: CFG-based data flow tracking with source-to-sink path explanation
• Symbol Context: 360° symbol view with incoming/outgoing references and disambiguation
• Rename Refactoring: Graph-backed multi-file coordinated renaming with confidence scoring
• Change Detection: SHA1 hash-driven incremental re-indexing
• Route / API Analysis: Framework-aware route extraction (Next.js, Laravel, Express, etc.), API impact analysis, and response shape checking
• Cross-Repo Groups: Multi-repo contract matching, bridge building, and cross-repo impact analysis
• Wiki Generation: LLM-augmented project documentation from the knowledge graph
• MCP Server: Model Context Protocol server for AI coding agent integration
• Embedding Pipeline: HTTP embedding with int8 quantization, mmap vector files, and two-stage search
• Scalability: Designed for 1M+ node repos — batched indexing, chunked BM25, sharded LRU cache, GC pools
• CLI: Full command-line interface with 20+ subcommands
• Go API: Embedded library API with extensible provider/resolver/tool registration

Changed

• Project definition established as Hybrid Graph-Augmented Code Intelligence Engine