text
| 1 | # Technical Writing Rules |
| 2 | |
| 3 | Agent prose has recognizable tells: an opening that restates the question, three |
| 4 | hedges per paragraph, a bulleted list where a sentence would do, and a closing offer |
| 5 | nobody asked for. These rules cut all of it, and add the one that matters most, which |
| 6 | is never sounding more certain than you are. |
| 7 | |
| 8 | ## When to use |
| 9 | |
| 10 | - Documentation, READMEs, commit messages, PR descriptions, and code comments. |
| 11 | - Status updates and summaries an agent writes for a human to read. |
| 12 | - Anywhere an agent's output will be read by someone who was not watching it work. |
| 13 | |
| 14 | ## Install |
| 15 | |
| 16 | ```bash |
| 17 | npx openagents-cli add openagents/writing-style-rules |
| 18 | ``` |
| 19 | |
| 20 | | Runtime | Installed to | |
| 21 | |---|---| |
| 22 | | `claude-code` | `.claude/skills/writing-style-rules/` | |
| 23 | | `cursor` | `.cursor/rules/writing-style-rules/` | |
| 24 | | `codex` | `.codex/skills/writing-style-rules/` | |
| 25 | | `openai-agents` | `agents/writing-style-rules/` | |
| 26 | | `generic` | `.openagents/writing-style-rules/` | |
| 27 | |
| 28 | ## What is in the package |
| 29 | |
| 30 | - `RULES.md` - the rules, with a before and after for each. |
| 31 | - `patterns.md` - the specific phrases to cut, and what to write instead. |
| 32 | |
| 33 | ## License |
| 34 | |
| 35 | MIT |
| 36 |