Skip to content

Releases: ShreyashChaurasia/CodexLens

CodexLens v0.2.1

Choose a tag to compare

@github-actions github-actions released this 21 Jul 22:38

CodexLens v0.2.1

CodexLens v0.2.1 is a compatibility patch for the production security-review
workflow: local static analysis, optional OpenAI business-logic review, and
human-approved patching.

Highlights

  • Replaced the Unicode banner used by root help and codexlens --version with
    an ASCII-only wordmark, so the command works in Windows terminals using a
    legacy code page.
  • Kept package metadata, lockfiles, the PyPI workflow default, and the owned
    ExpenseFlow example on the same 0.2.1 version.

Upgrade notes

  • codexlens --version remains human-readable branded output and is now safe
    to run in standard Windows consoles. Tools that need only the installed
    version should use Python package metadata.

Verification

This release is built from the v0.2.1 tag after locked dependency install,
Ruff checks, the full test suite, and source/wheel builds. The attached
artifacts are the release source distribution and universal Python wheel.

Getting started

See the README for
installation and the optional live OpenAI review workflow. Review the
security policy
before scanning code that may contain sensitive material.

CodexLens v0.2.0

Choose a tag to compare

@github-actions github-actions released this 21 Jul 22:13

CodexLens v0.2.0

CodexLens v0.2.0 focuses the project on its production security-review
workflow: local static analysis, optional OpenAI business-logic review, and
human-approved patching.

Highlights

  • Removed the legacy non-production walkthrough command and temporary project
    planning material.
  • Kept ExpenseFlow as a focused, intentionally vulnerable integration example
    with exploit proof, hardened-reference behavior, and a live patch regression
    path.
  • Added the ANSI Shadow CodexLens mark to the README, root help, and version
    output.
  • Retained JSON reporting, confirmation-gated patches, strict local patch
    validation, and model-neutral OpenAI Responses API support.

Upgrade notes

  • codexlens demo has been removed. Use codexlens scan for local static
    analysis or a configured --model review instead.
  • codexlens --version is now human-readable branded output. Tools that need
    only the installed version should use Python package metadata.

Verification

This release is built from the v0.2.0 tag after locked dependency install,
Ruff checks, the full test suite, and source/wheel builds. The attached
artifacts are the release source distribution and universal Python wheel.

Getting started

See the README for
installation and the optional live OpenAI review workflow. Review the
security policy
before scanning code that may contain sensitive material.

CodexLens v0.1.0

Choose a tag to compare

@github-actions github-actions released this 20 Jul 09:28

CodexLens v0.1.0

This is the first public release of CodexLens: a Python CLI security auditor
for reviews that need more than pattern matching.

Highlights

  • A three-pass workflow combines local static checks, optional AI-assisted
    business-logic review, and confirmation-gated patch proposals.
  • Any compatible OpenAI model can be selected with --model or
    CODEXLENS_MODEL; CodexLens does not hard-code a model family.
  • Proposed fixes are constrained to a reviewed source unit, shown as a local
    unified diff, and applied only when the user enters y.
  • The repository includes JSON reporting for automation and an owned
    ExpenseFlow IDOR demonstration with exploit and hardened-reference tests.

Verification

This release was built from the v0.1.0 tag after locked dependency install,
Ruff checks, the full test suite, and source/wheel builds. The attached
artifacts are the release source distribution and universal Python wheel.

Getting started

See the README for
installation, the credential-free offline replay, and the optional live
OpenAI evaluation path. Review the
security policy
before scanning code that may contain sensitive material.