Changelog¶

Release notes will be published with tagged releases.

[Unreleased]¶

feat: implement check_architectural_patterns MCP tool — replace NotImplementedError stub with 16 language-agnostic pattern detectors (singleton, factory, observer, repository, builder, command, mvc, middleware, prototype, decorator, facade, proxy, strategy, composite, chain-of-responsibility, iterator) covering GoF and architectural patterns across Python, Go, Rust, JavaScript, Java, and TypeScript via compiled regex without requiring a parse tree
ci: add rrt-checks job using repo-release-tools@v1.8.1 for branch name, commit subject, changelog, and release-health policy validation
ci: add build-mcpb job that produces a DXT v0.4 .mcpb bundle and attaches it to every GitHub Release alongside wheel, sdist, and SBOM
release: add scripts/build_mcpb.py and poe build_mcpb task to build the .mcpb bundle locally (dist/ is gitignored)
docs: document .mcpb one-click install for Claude Desktop in installation and MCP-integration pages
dev: add CLAUDE.md with project conventions for Claude Code sessions

release: .mcpb manifest schema corrected to DXT spec v0.4 (manifest_version, repository as object, server.entry_point added)
release: replace pyproject.toml entry-point with a stdlib-only server/main.py shim — bundling pyproject.toml triggered uv_build.build_editable, which failed without the full source tree
release: use uv tool run instead of uvx in mcp_config so the command is valid regardless of how Claude Desktop resolves the binary path
fix: update CHANGELOG.md with recent edits and ensure formatting consistency

ci: upload test coverage XML and artifact for codecov/upload (#156)
enhance MCP integration with Codex and Copilot support, update documentation and CLI commands (#149)

docs: repair Mermaid contrast, text overflow, and expand diagram coverage (#137)

Batch mode — LLM-safe chunked analysis with cursor pagination (#110)
Fix overall_score scale (0–100 → 0–10) to match documented contract in models.py (#110)
Implement proper weighted health score in analyze_batch_summary: severity² penalty (70%), violation density (20%), maintainability index (10%) (#19, #110)
Add analyze_batch_auto MCP tool: auto-routes to full result or paginated batch based on token budget (#20, #110)

enhance branch creation with normalization and description joining functions; add tests for dry-run behavior (#107)
Add support for Ansible, Terraform, and SVG analysis (#105)
complete zen rules for all 5 languages (#75–#79) (#95)

Add philosophy document and update links in README; adjust argument names in analyzers (#90)
enhance language analyzers with AST parsing capabilities (#71)

ci: enable verbose output for TestPyPI package publishing
gitlab_ci: Add GitLab CI support with multiple detectors and principles (#67)
enhance language validation and documentation generation (#65)

css: resolve back-to-top flip, contrast failures, and add cherry-red palette (#46)
linting issues ruff (#45)