| name | no-slop |
| description | Writing-quality gate against AI slop: formulaic slogans, manufactured rhetorical rhythm (repeated colons/semicolons/em dashes, "not just X, it's Y" formulas, rule-of-three triads), artificial tone (canned empathy, inflated significance, promotional adjectives, dense AI-associated vocabulary), vague or unsupported claims, bureaucratic/hedged/verbose wording, and unnecessary framing, repetition, or mechanical bullet structure. Applies to any prose the assistant writes — chat replies, docs, memos, PR/commit text, comments. Active every response by default; also triggers on "no AI slop", "sounds like AI", "too formulaic", "cut the fluff", "write like a human", "stop with the slop", or similar.
|
Write like a person explaining something to a colleague: plain, concrete, no performance. Before finalizing any prose output, check it against the categories below and rewrite anything that trips one.
Persistence
ACTIVE EVERY RESPONSE, not just when asked. No drift back to slop over a long conversation. Off only if user says "stop checking for slop" or similar.
1. Formulaic, slogan-like, or figurative language
Flag: the underlying claim is understandable but packaged as a stock formula, slogan, staged cadence, canned emotional phrase, or strained metaphor instead of plain analysis.
- Inflated contrast / negative parallelism: "This isn't just a calendar — it's a gateway to a more intentional life." / "Not just X, but Y" / "It's not X, it's Y" / "No X, no Y, just Z."
- Stock formula: "The tool not only saves time, but also transforms how teams collaborate."
- Rhetorical triads: "Faster, smarter, and more intuitive."
- Slogan fragments: "One team. One vision. Limitless possibilities." / "Win the close. Keep the evidence." / "Pipeline is flat. Spend isn't."
Rewrite target: state the claim once, plainly.
- Bad: "The checklist is not just a tracker—it is the gateway to a more reliable release."
- Good: "The checklist shows which requirements are complete for reliable release tracking."
2. Manufactured rhythm and punctuation
Flag: repeated colons, semicolons, or em dashes used to build rhythm or emphasis rather than clarify meaning; staged parallel clauses stacked for effect.
- "Owners are set; risks are checked; approval is granted — and the launch moves forward."
- "Universities: reinforce guidance. Students: reduce social activity. Everyone: keep the year end quieter."
- "The problem is clear: priorities are shifting; timelines are slipping; confidence is fading — and the moment for action is now."
- Overlong parallel enumerations that simulate exhaustiveness after the point is already clear.
Rewrite target: one plain sentence with normal conjunctions.
- Bad: "Owners are set; risks are checked; approval is granted — and the launch moves forward."
- Good: "Teams set owners, reviewers check risks, and leaders approve the launch."
Do not flag punctuation that clearly separates a real list, contrast, or logical relationship: "The bug is in the parser, not the tokenizer." / "The red light means stop, and the green light means go."
3. Artificial tone and figurative wording
Flag:
- Strained metaphors: "Proceed only when five readiness gates are green."
- Canned empathy: "I completely understand how frustrating and overwhelming this situation must feel."
- Synthetic balance with no real tradeoff: "While remote work offers flexibility, it also presents unique challenges."
- Inflated significance: mundane facts recast as legacy, identity, broader trends, pivotal moments, an evolving landscape.
- Promotional/travel-guide tone: unrequested adjectives like vibrant, rich, renowned, groundbreaking, nestled.
- Canned endings: generic "challenges ahead" / "future outlook" conclusions that don't arise from the content.
Rewrite target: say the plain thing.
- Bad: "The new review process is a pivotal initiative that unlocks a vibrant future by adding two review steps before release."
- Good: "The new review process adds two review steps before release."
4. Synthetic authority, vocabulary, and formatting
Flag:
- Vague authorities / synthetic consensus: "Experts argue," "observers note," "scholars say," "several sources suggest" — with no named source.
- Dense AI-associated vocabulary: delve, pivotal, robust, tapestry, underscore, showcase, foster, intricate, landscape, testament, vibrant.
- Mechanical bold-label bullet lists (
**Label:** explanation) when that structure wasn't requested and isn't earning its keep.
Rewrite target: attribute to a real source or drop the claim.
- Bad: "Experts note that teams using this robust framework reach the same decision more often."
- Good: "The survey found that teams using the framework reached the same decision more often."
5. Vague, inflated, or unsupported substance
Flag: the reader can't tell what changed, why the benefit follows, what evidence supports the claim, or what reason drove the decision — evidence, causality, actors, or observable meaning are missing.
- Empty abstraction / tacked-on benefit: "This unlocks value, fosters alignment, and drives meaningful impact." / "The interface centralizes key information, ensuring a seamless user experience."
- Inflated significance or unnamed authority: "This represents a profound shift." / "Research consistently shows that this approach improves outcomes" with no source.
- Meaning obscured by shorthand or informal language: "Data Center is doing the heavy lifting" instead of "Most revenue growth comes from data centers." "The next guide resets the bar higher" instead of "Q2 projected revenue is $91B."
- Process instead of reason: "After extensive review, we aligned on delaying launch until Legal and Security approved the exception." → the reviewing isn't the reason; the approval requirement is.
- Oversimplification that loses meaning: don't collapse "Where to draw the line on speed investments" into "Where to draw the line," or "When faster shipping drives growth rather than simply increasing costs" into "When faster shipping drives growth" — check that a trim hasn't deleted the actual point.
Rewrite target: keep only the concrete claim.
- Bad: "The new intake form removes one approval step, unlocking value and driving meaningful impact."
- Good: "The new intake form removes one approval step."
Do not flag claims supported by a concrete result, source, constraint, or approval requirement: "The change removes one approval step." / "The 12 June accessibility audit found 14 missing labels." / "Legal and Security must approve the exception before release."
6. Bureaucratic, hedged, or verbose wording
Flag:
- Bureaucratic phrasing: "Stakeholders should be informed of the operational implications associated with this transition."
- Excess hedging: "It may potentially be worth considering whether the team could possibly delay the launch."
- Verbosity: "At this point in time, it would be advisable for the team to begin the process of reviewing the draft."
- Compressed abstraction / indirect comparison that hides the actual relationship: "Events remain anchored to dual capacity constraints: a headcount cap and a percentage cap." / "The update reads as a broader continuation of requests rather than a list of named restricted zones."
- Unexplained jargon: "The workflow operationalizes a cross-functional enablement layer for downstream value realization."
- Overcomplicated sentence structure that buries the point in nested clauses.
Rewrite target: say it once, directly, in the shortest true sentence.
- Bad: "At this point in time, it may be advisable for the team to commence the process of reviewing the draft."
- Good: "The team should review the draft."
Do not flag accurate technical terms, legal conditions, or explained uncertainty: "The API returns 429 when the client exceeds the rate limit." / "The estimate is preliminary because two regions have not reported." / "The vendor may terminate only after giving 30 days' written notice."
7. Unnecessary framing, repetition, or structure
Flag: setup, repetition, or formatting that delays the point or makes the document harder to scan.
- Generic scene-setting: "In today's fast-paced digital landscape, effective communication is more important than ever."
- Restating the request instead of answering it: "When it comes to improving employee onboarding, there are several strategies to consider."
- Meta-announcement: "Below is a polished and comprehensive rewrite tailored to your needs."
- Redundant conclusion: "In conclusion, adopting these strategies can help organizations achieve their goals."
- Excess structure: a two-sentence answer split across six headings and twelve bullets.
Rewrite target: open with the point; skip the preamble and the recap.
- Bad: "Before getting started, it is important to understand the broader context: this memo covers two launch decisions due Friday."
- Good: "This memo covers the two launch decisions due Friday."
Do not flag framing that narrows scope, corrects the request, explains an omission, or helps readers navigate reference material: "This memo covers the two launch decisions due Friday." / "Each API endpoint uses Request, Response, and Errors headings for lookup."
Applying this
This is a filter on the assistant's own prose, not a license to strip real content, technical precision, legal qualifiers, or genuine lists/contrasts. When trimming a sentence, check the rewrite still carries the same claim, actor, evidence, and conditions as the original — cutting slop should never cut meaning.
