More screenshots

    Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+21 more

    Procedure Writing Rules: Steps Anyone Can Follow Cold

    1

    Runbooks written by the person who knows the system only work for that person. Your agent applies 12 rules from Simplified Technical English so a reader who wasn't there runs the steps right the first time.

    Free

    Security scanned
    Procedure Writing Rules: Steps Anyone Can Follow Cold

    Procedure Writing Rules: Steps Anyone Can Follow Cold

    Example session with this skill installed

    Rewrite this deploy runbook so someone who has never deployed this service can follow it.

    • Read your context and instructions
    • Compiled the procedure writing rules

    The agent splits each step that holds two actions, starts every step with a verb, moves the "if the health check fails, revert the tag" condition ahead of its action, moves a warning about the production database above the migration command, replaces "it" with the object's name across steps, uses one name for the service throughout instead of three, and adds the expected output after the steps that can fail silently. It returns the rewritten runbook, which is longer than the original.

    Connects securely to your tools. The creator never sees your data.

    About this skill

    About this skill

    Instructions written from inside a system skip articles, pack two decisions into one comma, use "it" to point back three steps, put the warning after the command it guards, and leave prohibitions as trailing fragments a skimming reader misses. The author reads them fine. The tired person on call does not.

    This skill applies 12 rules taken from ASD-STE-100, Simplified Technical English, the standard written in 1986 for aircraft maintenance manuals so a mechanic could not misread a step. One instruction per sentence. Start with the verb. At most 20 words per step. Use articles and complete sentences. One word for one thing, every time. No -ing forms, metaphors, or idioms. Condition before action, warning before the step. No pronouns across step boundaries. State the expected result of any step that can fail silently. Write prohibitions as full sentences. It applies only to text someone will execute, and stays out of chat, analysis, marketing copy, and commit messages.

    The proof

    There is no measurement yet, and the listing claims none. The SKILL.md includes three worked before-and-after rewrites (a build instruction, a publish step, and an access checklist step), each with the reason the original fails, so you can judge the rules on real text. It also states its own cost: procedures get roughly 30 to 50 percent longer. Its kill condition is written in: drop the skill if rewritten runbooks show no fewer clarifying questions or failed first runs after two months, measured with the author's skill-eval tool.

    What you get

    One SKILL.md with the scope (what it applies to and what it does not), the 12 rules, the three before-and-after examples, and what the rules cost. It takes the STE rules but not the 900-word controlled vocabulary, since no checker runs.

    How to use

    1. Add the skill folder to your agent's skills directory (for Claude Code, ~/.claude/skills/procedure-writing-rules/).
    2. Ask the agent to write or rewrite a runbook, deploy guide, checklist, or the steps section of a SKILL.md.
    3. Give the result to someone who has not done the task and see whether they finish it without asking a question.

    Source and license

    Free to use and MIT licensed. The full source, install steps, and issue tracker are on GitHub at https://github.com/robertmnyborg/everyday-agent-skills.

    Uses the STE writing rules only, not the controlled vocabulary, and runs no automated checker, so compliance depends on the agent. Procedures get longer, roughly 30 to 50 percent by the skill's own estimate. Not meant for persuasive or analytical writing. No measured results yet.

    How to use

    1. 1

      Connect your agent to the Agensi MCP

      Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup

    2. 2

      Sign in to Agensi

      Your agent shows a sign-in link. Sign in and click Allow.

    3. 3

      Ask for what you need

      Your agent loads this skill and uses it on your task. It always asks before any payment.

    Reviews

    No reviews yet

    Be one of the first to try it. Every listed skill goes through the trust checks below.

    Security scanned

    Checked by our 8-point scan before listing

    Fresh listing

    Recently published to Agensi

    Free forever

    No account required to browse

    Trust & safety

    Security scan

    Security scan passed 8 days ago

    • Free with an Agensi account

    Requires

    Any agent that loads SKILL.md files

    Permissions

    No special permissions declared or detected

    Listed8 days ago

    What's inside

    Frequently Asked Questions

    Agents write tidy numbered lists, but they still combine actions, put warnings after the step, and vary the name of the same thing. These 12 rules are specific enough to check line by line.

    Yes. It is one SKILL.md with no scripts, so any agent that reads SKILL.md can use it.

    No. It is scoped to procedural text only. Chat, analysis, strategy memos, marketing copy, commit messages, and PR descriptions are out of scope.

    No. The skill uses the writing rules only and keeps domain terms and identifiers as they are.

    On GitHub at https://github.com/robertmnyborg/everyday-agent-skills, under the MIT license. Every file in this package is there, plus the README and install steps.