| name | ste-linter |
| description | Lint prose against Simplified Technical English (ASD-STE100) rules. Use when the user asks to check, lint, or simplify technical writing (specs, manuals, procedures, docs) for STE compliance, or mentions ASD-STE100 or Simplified Technical English. |
Simplified Technical English Linter
Lint technical prose against a practical subset of the ASD-STE100 writing
rules: sentence length (20 words procedural / 25 descriptive), paragraph
length (6 sentences), active voice, simple present tense, no contractions,
no noun clusters over 3 nouns, and a controlled dictionary of unapproved
words and phrases with approved alternatives.
Running the linter
python3 ${CLAUDE_PLUGIN_ROOT}/scripts/ste_lint.py [--format text|json|feedback] <file.md ...>
- No dependencies; Python 3 stdlib only. Reads stdin when no files are given.
- Exit code is 1 when there are findings, 0 when clean.
--format text (default): one finding per line with file:line:col.--format json: machine-readable findings plus an error/warning summary.--format feedback: markdown grouped by rule — use this when handing the
findings to a writer (or another agent) to revise the document.
Workflow
- Run the linter with
--format feedback on the document.
- Revise the document to resolve each finding. Keep the meaning; change the
wording. Typical fixes:
- Split long sentences; one instruction or idea per sentence.
- Rewrite passive voice with the actor as the subject ("The API returns…").
- Replace "will/shall/should/may" with the present tense, "must", or "can".
- Substitute the suggested approved word ("utilize" → "use").
- Break noun clusters apart with prepositions.
- Re-run the linter until clean or until the remaining findings are false
positives (the checks are heuristic — judge each finding; do not contort
correct prose to silence a wrong one).
Caveats
This is an STE-inspired heuristic subset, not the official ASD-STE100
specification. The official spec and full dictionary are distributed by the
ASD STEMG (https://www.asd-ste100.org). Domain-specific technical names and
technical verbs are allowed by STE and may still be flagged here — treat
those findings as ignorable.
