editorial-standards
Skill14 starsChanged 8 months agoMITView on GitHub
The editing pass order, the measure-don't-assert rule, and which surface answers which question.
We split this skill into the rules it gives an agent
6 rules. About 0 are common, 6 are distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈6
Distinctive · estimate
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 28
Never say a draft "reads better" or "is much clearer".
- Distinctive: line 34
If you did not run the tool, you do not have the number.
- Distinctive: line 41
Never invent a fact, name, number, date, link, or quotation.
- Distinctive: line 42
If a claim looks wrong, unsupported, or unverifiable, flag it in a
- Distinctive: line 46
Keep the author's voice. You are not rewriting them into you.
- Distinctive: line 76
Every substantive edit returns, in this order:
≈ “Common” is an estimate: at least 1 in 5 of the 308,232 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Editorial Standards
This is the map, not the rulebook. It is inlined into your prompt on every turn, so it stays short. The full rulebook, the per-format checklists, and the preferred-term list live in the style-desk MCP server — query it when you need a specific rule.
Pass order
Edit in this order and never skip backwards mid-pass:
- Scope — what is it, who reads it, what must the reader do afterwards.
- Structure — outline, ordering, headings, missing and redundant sections.
- Line — sentences, verbs, hedges, length.
- Terminology — one name per concept, consistently spelled.
- Proof — checklist for the format, then a final read.
Reordering sections after line editing wastes the line editing. Structure first, always.
Measure, do not assert
Never say a draft "reads better" or "is much clearer". Show it:
readability_scorebefore and after, quoting both numbers.prose_lintbefore and after, quoting the finding counts.text_diffto show exactly what changed.
If you did not run the tool, you do not have the number. Do not estimate one.
The line you never cross
You may change how something is said. You may not change what it claims.
- Never invent a fact, name, number, date, link, or quotation.
- If a claim looks wrong, unsupported, or unverifiable, flag it in a Queries for the author list. Do not smooth it into something plausible.
- Preserve hedging that reflects genuine uncertainty; cut hedging that is only timidity. When you cannot tell which it is, ask.
- Keep the author's voice. You are not rewriting them into you.
Caveman mode is the one pass that changes form aggressively — and it is still
bound by this rule. It never runs unless the user asked for it, it never
touches code, commands, URLs, paths, numbers, or error strings, and
caveman_check errors rather than warns when any of those go missing.
Which surface answers which question
| Question | Use |
| Where are the weak sentences? | prose_lint |
| How hard is this to read? | readability_score |
| What exactly did my edit change? | text_diff |
| Is the document's skeleton sound? | heading_outline |
| Is this term spelled one way throughout? | term_consistency |
| Compress this to the bone (only if asked) | caveman_draft, then repair by hand |
| Did the compression lose anything load-bearing? | caveman_check |
| What is the house rule, and why? | lookup_style_rule (style-desk) |
| Is this draft shippable for its format? | get_checklist (style-desk) |
| Which spelling is house-preferred? | preferred_term (style-desk) |
| What does each caveman level do? | caveman_rules (style-desk) |
| When must I stop compressing? | caveman_guardrails (style-desk) |
Tool findings are evidence, not verdicts. A long sentence that carries a list may be correct; a passive construction is right when the actor genuinely does not matter. Suppress a rule with a stated reason rather than obeying it blindly.
Delivering an edit
Every substantive edit returns, in this order:
- The edited text, ready to copy.
- What changed — grouped by reason, not a line-by-line replay.
- Numbers — before/after readability and lint counts.
- Queries for the author — anything you could not resolve without them.
In this repository50 other files it gives its agents.
- packages/sdk/examples/repo-workspaces/seed/home/AGENTS.mdAGENTS.md
- .github/copilot-instructions.mdCopilot instructions
- agent-repair-loopSkill
- duroxide-orchestration-versioningSkill
- finance-research-standardsSkill
- valuation-methodsSkill
- pilotswarm-architectureSkill
- deployment-safetySkill
- incident-responseSkill
- add-activitySkill
- add-eventSkill
- add-testSkill
- add-toolSkill
- agent-versioningSkill
- azure-ps-test-runner-vmSkill
- debug-orchestrationSkill
- debug-testSkill
- horizondb-idiosyncrasiesSkill
- horizon-facts-eval-sweepSkill
- investigate-duroxide-runtimeSkill
- pilotswarm-aks-deploySkill
- pilotswarm-aks-resetSkill
- pilotswarm-new-env-deploySkill
- pilotswarm-perfSkill
- pilotswarm-portal-app-regSkill
- pilotswarm-portal-auth-assignmentsSkill
- pilotswarm-releaseSkill
- pilotswarm-tuiSkill
- pilotswarm-vpn-client-profileSkill
- schema-migrationSkill
- notesSkill
- share-a-fileSkill
- cost-latency-analysisSkill
- graph-debugSkill
- orchestration-session-lifecycleSkill
- resourcemgrSkill
- sweeperSkill
- canvas-appsSkill
- html-visualsSkill
- pilotswarm-agent-versioningSkill
- pilotswarm-aks-identitySkill
- pilotswarm-azure-deployerSkill
- pilotswarm-azure-lessonsSkill
- pilotswarm-cli-builderSkill
- pilotswarm-duroxide-versioningSkill
- pilotswarm-hybrid-datastoreSkill
- pilotswarm-knowledge-harvesterSkill
- pilotswarm-portal-builderSkill
- pilotswarm-sdk-builderSkill
- pilotswarm-three-tierSkill
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
Reviews can't be read right now.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

