Plain Writing helps you draft, edit, and audit prose that stays faithful to source meaning while removing AI slop, verbosity, jargon, and formatting excess.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install plain-writing
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install plain-writing using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is Plain Writing?
Plain Writing is an Openclaw Skills workflow for producing clear, reader-focused prose from drafts or source material. It is designed to preserve protected meaning, voice, and required terminology while removing clutter, unsupported claims, robotic structure, and unnecessary formatting.
It supports technical, workplace, long-form, and creative writing across three modes: strict, technical, and natural. The skill emphasizes evidence-aware rewriting, careful handling of uncertainty, and source fidelity, making Openclaw Skills useful for both high-stakes editing and everyday cleanup of prose.
Plain Writing Use Cases
- Rewrite AI-generated text into clearer, more human prose while preserving meaning.
- Edit technical or workplace writing to reduce jargon, repetition, and inflated claims.
- Audit drafts for unsupported assertions, weak causation, or hidden uncertainty.
- Clean up long-form content by removing recap-heavy structure and decorative headings.
- Transform source material into finished prose without inventing facts, sequence, or relationships.
- Refine regulated or controlled-language text while keeping exact terms, warnings, and requirements intact.
How Plain Writing Works
- Set the contract by identifying the operation, audience, purpose, genre, language mode, and required terms.
- Separate source content into protected items and content that can be questioned, narrowed, or cut.
- Preserve facts, names, values, quotations, warnings, risks, decisions, and other anchored material.
- Choose the appropriate language mode:
strict,technical, ornaturaldepending on the writing task. - Draft or rewrite the prose using direct language, clear structure, and one term per concept.
- Remove unsupported claims, filler, repetition, rhetorical framing, and AI-shaped structure.
- For edits and audits, compare the rewritten text against the source to ensure no protected meaning was lost or expanded.
- Run lint and review checks with the provided scripts when available, then revise only when changes preserve fidelity and genre.
- Return only the finished prose or the requested audit output, not a critique or report.
Plain Writing Setup
- Place the skill directory alongside its
references/andscripts/folders. - Review the guidance files before writing or editing:
cat references/ai-patterns.md
cat references/eval.md
- Prepare your draft or source text and save only candidate prose to a temporary file.
- Run the prose linter from the skill directory:
python3 <skill-dir>/scripts/lint_prose.py --mode technical /path/to/draft.md
- Use the correct mode for the task:
python3 <skill-dir>/scripts/lint_prose.py --mode strict /path/to/draft.md
python3 <skill-dir>/scripts/lint_prose.py --mode natural /path/to/draft.md
- For source-aware comparison, run the delta scorer:
python3 <skill-dir>/scripts/score_delta.py before.md after.md
- Add protected-token flags when exact names, values, identifiers, or terms must be preserved.
- If tooling is unavailable, perform the review manually and do not claim a script ran when it did not.
Plain Writing Data Schema & Taxonomy
The skill does not define a database schema. Instead, it organizes work around a protected-content ledger, a voice snapshot, and the final prose output.
| Artifact | Purpose | Notes |
|---|---|---|
| Protected-content ledger | Tracks facts and must-keep material | Separate rows for risks, mitigations, targets, owners, deadlines, proposed actions, and condition–consequence pairs |
| Voice snapshot | Records cadence and style constraints | Captures preferred words, formality, roughness, audience knowledge, and first-person usage |
| Draft / source text | Input material to rewrite, edit, or audit | Must be separated into protected and non-protected content where relevant |
| Final prose | Completed output | Return only the requested prose, not commentary or a report |
| Lint results | Review evidence from lint_prose.py |
Warnings are review questions, not automatic failures |
| Score delta output | Source-aware comparison from score_delta.py |
Helps verify structural and protected-content fidelity |
Content taxonomy emphasized by Openclaw Skills in this skill:
- Protected: facts, names, values, examples, dates, quotations, sources, warnings, limits, uncertainties, and commitments
- Reviewable: unsupported claims, repetition, decorative structure, rhetorical synthesis, and generic filler
- Mode-specific:
strict,technical, andnaturalcontrol how aggressively prose is constrained
Plain Writing Advanced Features
- Three-mode writing control with
strict,technical, andnaturalsettings for different prose contexts. - Protected-content preservation workflow that prevents invented facts, relationships, timing, or mechanisms.
- Source-aware audit and rewrite process that distinguishes supported text from unsupported interpretation.
- Linting integration via
scripts/lint_prose.pyfor fast review feedback and mode-specific checks. - Delta scoring via
scripts/score_delta.pyto compare before/after drafts and catch structural expansion. - Supports regulated, technical, workplace, and creative writing without forcing a one-size-fits-all tone.
- Built for Openclaw Skills users who need cleaner prose without losing attribution, uncertainty, or voice.
SKILL.md
Loading
Related Openclaw Skills
find-ebooks / 搜好书
find-ebooks is an Openclaw Skills tool that searches ebooks across Anna's Archive and major Chinese book platforms, returning rich metadata and copyable links.
Weekly Dev Tips
Weekly Dev Tips is a compact Openclaw Skills knowledge base that helps developers improve shell productivity, Git hygiene, and debugging judgment one actionable tip at a time.
ArcGIS Online
Openclaw Skills for ArcGIS Online lets AI agents safely search, geocode, and read ArcGIS data through a schema-first CLI connector.
Memos Connector Skill
Openclaw Skills for Memos provides a secure CLI-driven way to read, create, update, and delete Memos data through a connected OOMOL account.
WeChat Article AI Check
A WeChat article review skill that checks originality, AI-like risk, and structural quality before guiding a safer rewrite.
iWork to Markdown
iWork to Markdown is an Openclaw Skills parser that converts Apple Pages, Numbers, and Keynote files into clean Markdown text.
Featured*
