# LintLang 0.8.2

Published **2026-09-30** · Stable release · [GitHub release](https://github.com/hermes-labs-ai/lintlang/releases/tag/v0.8.2) · [PyPI](https://pypi.org/project/lintlang/0.8.2/)

## [0.8.2] - 2026-09-30

### Added

- A Simplified-Chinese quickstart (`docs/zh-CN/`) with a drift guard keeping
  it in step with the English docs. Added in
  [#156](https://github.com/hermes-labs-ai/lintlang/pull/156).

### Changed

- **Breaking:** the GitHub Action's `fail-on` input now defaults to empty,
  making the workflow advisory: it reports findings but only fails on input
  errors, matching the CLI and pre-commit hook defaults. Workflows that relied
  on the previous blocking default must set `fail-on: fail` (or
  `fail-on: review`) explicitly to keep enforcing; the change takes effect when
  you move the `uses:` ref to this release. The Action only passes `--fail-on`
  to the CLI when the input is non-empty, and the input is now
  whitespace-trimmed. Shipped in
  [#149](https://github.com/hermes-labs-ai/lintlang/pull/149).
- `skills/lintlang/SKILL.md` (previously the Pi-only skill) is now a
  host-neutral agent skill: it resolves a runner — installed CLI first, then a
  pinned `uvx` release (fetched on a cache miss), then a pip install line —
  and walks through auditing an instruction file with the deterministic
  scanner. Verified against this release. Replaced in
  [#154](https://github.com/hermes-labs-ai/lintlang/pull/154).

### Fixed

- Scanning a UTF-16 or otherwise non-UTF-8 file no longer fails with a bare
  `UnicodeDecodeError`: the message now says LintLang reads UTF-8, names the
  detected encoding shape, and tells you to save or convert the file as UTF-8.
  The improved decoding is shared by the file and standard-input paths, so
  piped input gets the same hint instead of a generic read failure. Reported by
  [@roli-lpci](https://github.com/roli-lpci) in
  [#151](https://github.com/hermes-labs-ai/lintlang/issues/151) (labeled
  `good first issue`); fixed by [@flypov](https://github.com/flypov) in
  [#153](https://github.com/hermes-labs-ai/lintlang/pull/153).
- `lintlang scan --fail-under` now validates the threshold as a finite float
  in the inclusive range `[0, 100]` before scanning. Previously `nan`,
  negative, and `-inf` values were accepted and silently disabled the score
  gate (exit 0 on a clean fixture), while `inf` and values above 100 made the
  gate always fail with a confusing "below threshold" message. Invalid values
  now exit 2 with a clear argparse error. Reported in
  [#150](https://github.com/hermes-labs-ai/lintlang/issues/150); fixed in
  [#160](https://github.com/hermes-labs-ai/lintlang/pull/160).
- The Claude Code plugin bundle manifests track the scanner release:
  manifest versions are asserted equal to `lintlang.__version__` by a new
  regression test, so a bundle can no longer silently ship a stale version.
  The bundle also moved the hook's runner check from an exact `==` pin to a
  `>=` minimum-version gate, so a deliberately newer installed scanner is
  accepted instead of refused. Fixed in [#148](https://github.com/hermes-labs-ai/lintlang/pull/148).

## [0.8.1] - 2026-09-29


## Upgrade

```sh
python -m pip install --upgrade lintlang==0.8.2
```

## Compatibility and limitations

See the published release notes above for compatibility and migration details. Host integrations may carry their own tested pins.

[GitLab setup](/integrations/gitlab) · [Outputs and exits](/docs/outputs) · [Quickstart](/docs/quickstart) · [Complete changelog](/releases/changelog).