writing-quality
Skill98 starsChanged 25 days agoMITView on GitHub
Removes common AI writing patterns while preserving meaning, evidence, structure, and SDDP traceability. Ambient through AGENTS.md; not directly invokable.
9 other projects use a version of this.
Adopted by
Most recently taken up by
- shawnclybor/clybor-claude-toolingtheir own version, first seen Nov 8, 2026
- getsentry/software-factorytheir own version, first seen Oct 25, 2026
- vercel-labs/sanity-copilot-eve-templatetheir own version, first seen Oct 13, 2026
- vercel-labs/typefully-eve-templatetheir own version, first seen Oct 13, 2026
- vercel-labs/kody-eve-templatetheir own version, first seen Oct 13, 2026
- vercel-labs/eve-software-factory-templatetheir own version, first seen Oct 13, 2026
Other versions9 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 9
- vercel-labs/eve-software-factory-template★ 1.2ksame name
- vercel-labs/kody-eve-template★ 13same name
- vercel-labs/typefully-eve-template★ 8same name
- shawnclybor/clybor-claude-tooling★ 6same name
- vercel-labs/sanity-copilot-eve-template★ 5same name
- guerrilla2799/event-and-scale-os★ 2same name
- guerrilla2799/ops-and-scale-os★ 1same name
- ItalyPaleAle/skills★ 1same name
- getsentry/software-factory★ 0same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository73 other files it gives its agents.
- AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- .github/copilot-instructions.mdCopilot instructions
- sddp-amendSkill
- sddp-analyzeSkill
- sddp-autopilotSkill
- sddp-checklistSkill
- sddp-clarifySkill
- sddp-devopsSkill
- sddp-devsetupSkill
- sddp-implement-qc-loopSkill
- sddp-implementSkill
- sddp-initSkill
- sddp-planSkill
- sddp-prdSkill
- sddp-projectplanSkill
- sddp-qcSkill
- sddp-regenSkill
- sddp-reverseengineerSkill
- sddp-specifySkill
- sddp-systemdesignSkill
- sddp-tasksSkill
- sddp-amendSkill
- sddp-analyzeSkill
- sddp-autopilotSkill
- sddp-checklistSkill
- sddp-clarifySkill
- sddp-devopsSkill
- sddp-devsetupSkill
- sddp-implement-qc-loopSkill
- sddp-implementSkill
- sddp-initSkill
- sddp-planSkill
- sddp-prdSkill
- sddp-projectplanSkill
- sddp-qcSkill
- sddp-regenSkill
- sddp-reverseengineerSkill
- sddp-specifySkill
- sddp-systemdesignSkill
- sddp-tasksSkill
- adr-authoringSkill
- artifact-conventionsSkill
- clarification-strategiesSkill
- compact-communicationSkill
- evidence-led-authoringSkill
- implementation-standardsSkill
- instructions-managementSkill
- markdown-compressionSkill
- plan-authoringSkill
- quality-assuranceSkill
- sdd-methodologySkill
- sddp-amendSkill
- sddp-analyzeSkill
- sddp-autopilotSkill
- sddp-checklistSkill
- sddp-clarifySkill
- sddp-devopsSkill
- sddp-devsetupSkill
- sddp-implement-qc-loopSkill
- The first 60 shown.
The file
Show its rules and text
We split this skill into the rules it gives an agent
21 rules. 0 are common, 18 are distinctive, and 3 are very specific to this project.
0
Common
Most of the 9 edited versions we read keep these.
18
Distinctive
This author's own choices. Worth a look.
3
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 22
Add human judgment where the format allows it.
- Distinctive: line 23
Ask, "What makes this obviously AI generated?" Fix any remaining tells.
- Distinctive: line 24
Run the owning structural validator after the writing pass.
- Distinctive: line 28
Never run a whole-file style rewrite over SDDP artifacts or governance files.
- Very specific: line 40
Do not use writing quality to resolve [NEEDS CLARIFICATION], alter acceptance criteria, reorder priorities, or mutate artifact grammars.
- Distinctive: line 42
If deterministic Markdown compression also applies, use this order:
- Distinctive: line 45
Run validator-backed Markdown compression.
- Distinctive: line 46
Run the artifact's structural validator.
- Very specific: line 53
Use I when first person fits the speaker and format.
- Distinctive: line 54
Prefer concrete facts, mechanisms, examples, and measured values.
- Distinctive: line 55
Keep natural structure. Do not force every point into a polished three-part list.
- Distinctive: line 78
Avoid em dashes in prose. End the sentence or use a comma.
- Distinctive: line 79
Use colons for lists or examples, not as routine sentence connectors.
- Distinctive: line 80
Do not bold every proper noun, acronym, or list label.
- Distinctive: line 81
Avoid inline-header lists that repeat their labels.
- Distinctive: line 82
Use sentence case for newly authored narrative headings unless a required template fixes the heading text.
- Distinctive: line 83
Remove decorative emoji and use straight quotes in newly authored prose.
- Distinctive: line 87
Remove chatbot phrases such as "I hope this helps", "Let me know if", "Of course", and "Certainly".
- Distinctive: line 89
Remove empty disclaimers. Find the missing source or omit the unsupported claim.
- Distinctive: line 98
Prefer active voice when the actor matters and is known.
- Very specific: line 100
Prefer plain words: use, help, many, and if instead of utilize, facilitate, numerous, and in the event that.
“Common” is measured: kept by at least 6 in 10 of the 9 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Writing quality
Runtime primer:
AGENTS.mdsectionCommunication StyleunderWriting Quality. This file expands the editing rules and safety limits. It is not a required read during ordinary workflow execution.
Adapted from the user-provided Unslop rules.
Purpose
Edit prose so it sounds direct, specific, and written for this project. Remove stock AI phrasing without making the text sterile or changing what it claims.
Preserve meaning, scope, certainty, evidence, citations, and the user's voice. Style work never authorizes a new requirement, stronger claim, resolved ambiguity, changed priority, or removed caveat.
Process
- Scan newly written or changed prose for the patterns below.
- Rewrite only the affected narrative spans. Match the intended tone.
- Add human judgment where the format allows it. Be specific, vary sentence length, and use first person when it fits.
- Ask, "What makes this obviously AI generated?" Fix any remaining tells.
- Run the owning structural validator after the writing pass.
Safety limits
Never run a whole-file style rewrite over SDDP artifacts or governance files. In parser-sensitive files, edit only narrative spans created or changed by the current task.
Preserve these elements exactly unless the workflow explicitly owns their change:
- frontmatter, headings, required section order, tables, and column schemas
- checkbox lines and state
- IDs, priorities, markers, tags, paths, commands, URLs, and citations
- code fences, inline code, Mermaid, JSON, JSONL, YAML, OpenAPI, and GraphQL
- quoted user or source text
- uncertainty and legal, security, policy, or compliance language
- append-only history in
autopilot-log.mdanddivergence-log.md
Do not use writing quality to resolve [NEEDS CLARIFICATION], alter acceptance criteria, reorder priorities, or mutate artifact grammars. ADR files remain writable only through the ADR Author.
If deterministic Markdown compression also applies, use this order:
- Complete the writing-quality pass.
- Run validator-backed Markdown compression.
- Run the artifact's structural validator.
Voice
- State an opinion when the document calls for judgment. Do not manufacture one for factual or machine-readable output.
- Vary rhythm. Mix short sentences with longer ones that need the room.
- Acknowledge real tradeoffs instead of flattening them into neutral lists.
- Use
Iwhen first person fits the speaker and format. - Prefer concrete facts, mechanisms, examples, and measured values.
- Keep natural structure. Do not force every point into a polished three-part list.
Patterns to remove
Content
- Puffery such as "pivotal moment", "testament to", "evolving landscape", or "setting the stage". State what happened.
- Promotional words such as "vibrant", "breathtaking", "groundbreaking", "renowned", or "must-visit". Use a factual description.
- Vague attribution such as "experts believe" or "reports suggest". Name the source or remove the claim.
- Name-dropping without context. Keep the source that matters and state what it reported.
- Superficial
-ingclauses such as "highlighting", "showcasing", or "fostering". Delete them or explain the concrete effect. - Formulaic challenge language such as "despite challenges, it continues to thrive". Name the problem and result.
Language
- Stock words such as "additionally", "crucial", "delve", "enhance", "interplay", "intricate", "pivotal", "showcase", "tapestry", "testament", "underscore", and abstract "landscape". Use plain words.
- Inflated substitutes for
isorhas, including "serves as", "stands as", "boasts", and "features". - "Not just X, but Y." State the point directly.
- Forced groups of three, synonym cycling, and false "from X to Y" ranges.
- Technical-sounding metaphors such as "substrate", "wedge", "vector", "locus", "nexus", noun "primitive", metaphorical "harness", "surface", "bedrock", "scaffolding", "paradigm", "gold-plating", "ratchet", "evacuate", "endgame", "north star", and "flywheel". Name the actual component, action, or limit.
Style
- Avoid em dashes in prose. End the sentence or use a comma. Do not replace the dash with parentheses or another dash used for the same purpose.
- Use colons for lists or examples, not as routine sentence connectors.
- Do not bold every proper noun, acronym, or list label.
- Avoid inline-header lists that repeat their labels. Keep a lead-in only when the following text adds new information.
- Use sentence case for newly authored narrative headings unless a required template fixes the heading text.
- Remove decorative emoji and use straight quotes in newly authored prose.
Communication
- Remove chatbot phrases such as "I hope this helps", "Let me know if", "Of course", and "Certainly".
- Answer directly. Do not praise the question or agree with the user before giving the answer.
- Remove empty disclaimers. Find the missing source or omit the unsupported claim.
Plain speech
- Cut filler. Use "to" instead of "in order to" and "because" instead of "due to the fact that".
- Reduce stacked hedges to the level of uncertainty the evidence supports.
- Replace generic conclusions with a specific fact, decision, or next action.
- Say what a mechanism does. If a sentence could appear unchanged in another project's docs, make it project-specific or cut it.
- Split dense sentences. Prefer one idea per sentence when clauses make the reader backtrack.
- Prefer active voice when the actor matters and is known.
- Cut adverbs that prop up weak verbs. Use a stronger verb or a measured value.
- Prefer plain words:
use,help,many, andifinstead ofutilize,facilitate,numerous, andin the event that.
Output check
The pass succeeds when the prose is concrete, natural, and no less accurate than the source. If a cleaner sentence changes meaning or weakens a required safeguard, keep the original meaning and rewrite again.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
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.

