Slopless
Use this skill when asked to review English Markdown prose for AI slop, human slop, weak phrasing, cliches, filler, formulaic prose, or readability problems.
Scope
- Slopless is English-only.
- Slopless emits JSON only.
- Slopless reports findings. It does not rewrite prose.
- Do not use Slopless as a fact checker.
Required Workflow
- Run
npx slopless --helpbefore the first Slopless run in the session. - Create
.slopless/findingsin the current working directory. - Run Slopless on the requested Markdown files, folders, globs, or stdin.
- Save raw JSON output under
.slopless/findings/. - Use a timestamped filename that identifies the input.
- If the user asks for explanation, summarize the saved JSON findings for the user.
- Do not leave the only useful result in a temp directory.
Commands
mkdir -p .slopless/findings
npx slopless "docs/**/*.md" > ".slopless/findings/$(date +%Y-%m-%d-%H%M%S)--docs-review.json"
mkdir -p .slopless/findings
npx slopless draft.md > ".slopless/findings/$(date +%Y-%m-%d-%H%M%S)--draft.json"
mkdir -p .slopless/findings
npx slopless --stdin --stdin-filename draft.md > ".slopless/findings/$(date +%Y-%m-%d-%H%M%S)--stdin-draft.json"
Output Handling
- Exit
0: no findings. - Exit
1: findings were reported. - Exit
2: command failed before linting. - Treat exit
1as successful execution with prose findings. - Read the JSON before explaining results.
- Preserve rule IDs, file paths, line numbers, and excerpts when summarizing.
Ignore Rules
Use textlint comments when a finding is intentionally ignored:
<!-- textlint-disable slopless/semantic-thinness -->
Something shifted in the room.
<!-- textlint-enable slopless/semantic-thinness -->