CodeClaw v0.4.1

CodeClaw 0.4.1

Stable hardening release focused on packaging, cross-platform reliability, privacy workflows, and dataset lifecycle quality.

Highlights

• Added codeclaw diff to preview exactly what redaction pipeline will remove before confirmation.
• Added experimental codeclaw finetune --experimental scaffold for local Unsloth/QLoRA research workflows.
• Expanded classifier depth with weighted trajectory signal scoring.
• Added adapter-driven external source discovery and diagnostics for Cursor, Windsurf, Aider, Continue.dev, Antigravity, VS Code, Zed, and Xcode beta (experimental tier).
• Added encryption-at-rest primitives and setup/config hooks with key management status checks.
• Added immutable dataset version manifests + latest pointer metadata and publish dedupe index updates.
• Extended doctor/stats payloads with platform/adapters/encryption/versioning analytics.

CI and Distribution

• Added macOS tests in CI matrix (Linux + Windows + macOS).
• Added release build gates: python -m build, wine check, and CLI smoke checks.
• Added GitHub Packages publish step alongside PyPI flow.
• Added opt-in GitHub workflow template for team export automation on merged PRs.

Docs and DX

• Updated README and skill docs for setup/projects/diff/encryption/versioning workflow.
• Bumped package version to

CodeClaw v0.4.0-preview.5

Highlights

• Added guided onboarding via \codeclaw setup\ with source selection, dataset repo normalization, connected-project setup, MCP install, and optional watcher start.
• Added connected project management via \codeclaw projects\ (connect, disconnect, current-only, all, clear).
• Hardened Windows behavior for watcher triggering with safe fallback when \SIGUSR1\ is unavailable.
• Improved robustness for discovery and diagnostics paths in \doctor\, \stats\, and \share\.

Reliability and Tooling

• Added regression coverage for setup/projects flows, doctor failure handling, and Windows watcher fallback.
• Expanded CI to run tests on both Linux and Windows.
• Updated docs to reflect guided setup and connected-project workflow.

CodeClaw v0.4.0-preview.4

Highlights

• Public README finalized with no public roadmap section.
• MIT license metadata updated for 2025-2026.
• MCP memory UX milestone completed and covered by regression tests.
• Repo-wide branding drift checks added to prevent legacy naming regressions.
• Release docs clarified for trusted publisher troubleshooting.

Validation

• python -m pytest -q -> 369 passed
• python -m build -> sdist + wheel built

Artifacts

• codeclaw-0.4.0.tar.gz
• codeclaw-0.4.0-py3-none-any.whl

CodeClaw v0.4.0-preview.3

Highlights

• Stabilized v0.4 baseline with coherent CLI/docs naming and command consistency checks.
• Added MCP memory UX upgrades: cached index service, find_similar_sessions, and refresh_index.
• Added python -m codeclaw entrypoint for non-installed/dev usage.
• Hardened MCP install behavior for existing/corrupt ~/.claude/mcp.json scenarios.
• Expanded regression coverage for docs, markers, MCP tool behavior, and branding drift.

Validation

• python -m pytest -q (369 passing)
• python -m build (sdist + wheel)

Artifacts

• codeclaw-0.4.0.tar.gz
• codeclaw-0.4.0-py3-none-any.whl

CodeClaw v0.4.0-preview.2

CodeClaw v0.4.0 Preview 2

This preview release establishes the clean baseline for the new ychampion/codeclaw repository and ships the latest v0.4 feature set.

Added

• python -m codeclaw module entrypoint support.
• MCP tools for operational memory workflows:
  • find_similar_sessions(context, max_results)
  • refresh_index()
• Regression suite for documentation consistency and MCP behavior edge cases.

Changed

• Unified MCP tool response structure (ok, results/error, meta) for predictable integration behavior.
• Refactored MCP server internals to use a cached session/index service for stable query performance.
• Hardened install-mcp behavior to preserve existing MCP server entries and back up corrupt config files.
• Standardized command/documentation references to codeclaw and ---CODECLAW_JSON---.
• Updated CI to run docs consistency checks before the full suite.

Validation

• Full test suite: python -m pytest -q (366 passing).
• Build artifacts generated successfully:
  • codeclaw-0.4.0-py3-none-any.whl
  • codeclaw-0.4.0.tar.gz