Memory Lint

Use the version-pinned adapter to run memory-lint as a read-only diagnostic. It preserves the CLI's output and exit status and never installs dependencies while running.

Install

Install the exact supported release into the Python environment that will run the adapter:

python -m pip install memory-lint==0.1.0

Workflow

1. Locate the corpus configuration file. Every input is a named flag; do not pass a positional corpus path.
2. Prefer JSON for machine parsing and table output for a person reading the command directly.
3. Use --now YYYY-MM-DD when results must be reproducible across runs.

4. Run the adapter from this skill directory:

   python scripts/memory_lint_wrapper.py \
     --config path/to/config.yaml \
     --format json \
     --now YYYY-MM-DD

5. Interpret the exit status before reporting results:

   • 0: clean corpus
   • 1: one or more findings
   • 2: usage, configuration, filesystem, Git, dependency, or abnormal delegated-process error

An exit status of 1 is a successful lint run with findings, not an execution failure. Parse JSON only when the command produced valid JSON.

Corpus and Revision Options

• Use --corpus-root <path> to override corpus_root from the configuration without editing the file.
• Use --against <git-ref> only when the corpus is inside a Git repository and the ref is available locally. The CLI performs read-only Git operations.
• Supply --compare-before and --compare-after together to detect identical or whitespace-only revisions.
• Run python scripts/memory_lint_wrapper.py --help for the complete flag surface.

Report

Include the executed command, configuration and corpus paths, exit status, finding count, and findings grouped by severity and code. Report exit-2 errors verbatim enough to make the next action clear. Do not edit or auto-fix corpus files on the strength of lint output alone.