LintLangPi #
Give Pi a portable skill for reviewing named agent-instruction, configuration, and prompt files with your installed LintLang scanner.
Setup #
Use an installed Pi host with skill support, Git, and Python 3.10+. Install LintLang using pipx or pip, then confirm it is available:
command -v lintlang
lintlang --version
Open a terminal in an empty scratch directory and clone the released skill source:
git clone --depth 1 --branch v0.8.0 https://github.com/hermes-labs-ai/lintlang.git lintlang-integration-source
cd lintlang-integration-source
Install the released skill for personal use:
mkdir -p ~/.pi/agent/skills/lintlang
test ! -e ~/.pi/agent/skills/lintlang/SKILL.md && cp skills/lintlang/SKILL.md ~/.pi/agent/skills/lintlang/SKILL.md
The copy runs only if no skill file already exists; review an existing installation before replacing it. For project-only use, copy the same source file to .pi/skills/lintlang/SKILL.md inside your own project instead. Start Pi in your project, rather than the scratch checkout. Pi's skill discovery guide describes both locations.
Verify #
Ask Pi to audit a specific file, such as AGENTS.md. Confirm that it invokes the installed scanner using this form:
lintlang scan --format json -- AGENTS.md
The skill reports the verdict, rule codes, source locations, and repair suggestions. It distinguishes an input error or skipped file from a clean scan. It does not silently expand a named-file audit into a whole-repository scan.
Choose the workflow #
FAIL and REVIEW findings are advisory unless you request a gate. The skill's default requested gate is --fail-on review for MEDIUM and above; you can explicitly choose --fail-on fail for HIGH and above. There is no automatic edit hook in this integration.
For a missing skill, check its discovery path in your Pi installation. For a missing runner, check PATH in the host environment. Remove only the copied LintLang skill directory to undo installation.