Claude

Verified against Claude · 2026-08-06

Define a Custom Style once so every future chat gets the same tone by default

A prompt for authoring a reusable Claude Custom Style — concrete, directly actionable patterns instead of vague adjectives, explicit anti-patterns, and a before/after demonstration — so a saved formatting and tone preference actually holds up across chats instead of drifting back to Claude's own default.

Claude (Custom Styles)Claude.ai5 fillable variables

The prompt

Ready to copy — highlighted parts are example details you can swap.

Write a Custom Style definition I can save in Claude's settings, so this formatting and tone applies automatically to future chats without me re-explaining it or pasting samples each time. A saved Style is a short standing description Claude reads before every reply, not a one-off instruction, so word choice here matters more than in a normal prompt — vague adjectives left in a saved Style get applied vaguely forever, not just once.

WHAT THIS STYLE IS FOR
Internal Slack-style status updates to a technical team that reads a lot of these and has no patience for preamble.

CONCRETE PATTERNS TO ENCODE
Lead with the outcome or blocker in the first sentence. Use short paragraphs, no more than 3 sentences. State numbers and dates exactly, never 'recently' or 'soon.' End with a clear next action or 'no action needed' if there genuinely isn't one. State these as instructions Claude can act on directly ("open with the direct answer before any context," "never use more than one exclamation point per response, and only when something genuinely surprising happened"), not as adjectives ("be punchy") that leave the actual behavior underspecified.

WHAT THIS STYLE SHOULD AVOID
No 'I hope this finds you well' or similar opener. No hedging phrases like 'just wanted to check in' before the actual point. No exclamation points. No emoji. Naming the specific thing to avoid matters as much as naming what to do — a Style that only says what to do leaves Claude free to drift back toward whatever its own default happens to be for everything it doesn't explicitly cover.

BEFORE / AFTER EXAMPLE
Write one short example (2-4 sentences) in Claude's typical unstyled default voice, then rewrite the same content applying every rule above. The contrast should make each rule's effect visible and checkable, not just asserted — a Style description that never gets demonstrated against a concrete example is much easier to write vaguely without noticing.

CONSTRAINTS
- Keep the whole Style description itself under 150 words for the Style description itself — a Style that is too long to actually be read as a standing rule defeats its own purpose, and the discipline of fitting the pattern into a tight description is itself a good check on whether the pattern is concrete enough to state briefly.
- Do not describe a Style so specific it would look wrong applied to a very different kind of request — if someone pastes an unrelated creative writing request into a chat with this Style active, the output should just be plain, direct prose — not forced into a fake status-update shape — a Style meant for Internal Slack-style status updates to a technical team that reads a lot of these and has no patience for preamble. should still produce something sensible if applied to an unrelated question, even if it's not the ideal use case, rather than producing broken or bizarre output outside its intended lane.

IF I ALREADY HAVE ANOTHER SAVED STYLE FOR A DIFFERENT PURPOSE
Do not try to reconcile the two into one universal style — a Style is meant to be switched between per chat, not merged, so keep this one narrowly true to Internal Slack-style status updates to a technical team that reads a lot of these and has no patience for preamble. even if it means having more than one saved Style for different situations.

OUTPUT
The Style description, ready to paste into Custom Style settings as-is, followed by the before/after example clearly labeled as a demonstration, not part of the saved Style text itself.

Customize

Optional — swap in your own details for the highlighted parts above.

Why this works

A saved Style is read before every future reply, unlike a one-off prompt, which changes the cost-benefit of vagueness: an adjective-only instruction like 'be punchy' in a single chat gets one shot to be interpreted reasonably, while the same vague word saved as a standing Style gets applied, and drifts, across every future chat indefinitely — which is why this prompt insists on directly actionable phrasing over adjectives specifically for this feature, more than it would matter for an ordinary prompt used once. Naming anti-patterns as explicitly as patterns matters because a Style description that only lists desired behaviors leaves every unaddressed dimension to whatever Claude's own baseline default already is, and that baseline — measured, hedge-friendly, moderately warm — is precisely what most Custom Styles exist to override; without an explicit 'and never do X,' the model has no signal that a specific default behavior needs actively suppressing rather than simply not being mentioned. The before/after demonstration requirement makes a vague rule visible as vague before it gets saved: writing the same content twice, once unstyled and once styled, forces every claimed pattern to actually show up as a visible difference between the two versions, and a rule that produces no visible difference when applied has been caught as underspecified before it's committed to a standing setting that's much harder to notice is failing once it's just quietly running in the background of every future chat. The out-of-scope sanity check targets a risk specific to saved Styles versus a one-off prompt: a Style is active across every chat regardless of topic, so a Style tuned tightly for one narrow use case can produce genuinely broken output on an unrelated request if it was never checked against anything outside its intended lane — a failure mode a single-use prompt structurally cannot have, since it only ever applies to the one request it was written for in the first place.

Verified against

Claude Sonnet 4.6 (Custom Styles) · 2026-08-06

Changelog

  • 2026-08-06 — Initial publish, verified against Claude Sonnet 4.6 Custom Styles.

Building this for real?

This is a free starting point. If you'd rather have what Scult builds built and running for your business, that's Scult's day job.

EXPLORE WHAT SCULT BUILDS
Set up a Claude Project that stays accurate for months, not just this chatA structured Claude Projects setup that splits durable knowledge from custom instructions and adds an explicit staleness-check protocol, so every new chat in the Project inherits accurate context automatically instead of the knowledge base quietly rotting into confidently wrong answers.Claude (Projects)Claude Enterprise2026-07-20Build a multi-view dashboard as one self-contained Claude ArtifactA prompt for building a single-file interactive dashboard Artifact with multiple tabs sharing one dataset and one state model, so filters and selections stay consistent across views instead of several disconnected mini-tools bolted together.Claude (Artifacts)Claude.ai2026-07-21Audit a full contract for risk using Claude's whole context window, not a skimA long-context prompt that forces a clause-by-clause, quote-grounded risk audit of an entire pasted contract or policy document, ranked by severity and tied to one party's actual position, instead of a general summary of what the document is about.Claude (Opus 4.6, 1M context)Claude (Sonnet 4.6, 200K context)2026-07-22Feed Claude untrusted or user-submitted text without letting it hijack your instructionsAn XML-tagged prompt structure for any task where Claude must process content you did not write yourself — customer submissions, scraped pages, forum posts — that keeps instructions and untrusted data in separate labeled blocks and requires flagging any embedded instruction-injection attempt.Claude (Sonnet 4.6)Claude (Opus 4.6)2026-07-23
All Claude prompts

Check your AI visibility

One URL in, a 0–100 score and the exact fixes out.

RUN THE CHECK

Browse all the tools

15 tools across six categories
13 of them never send your data anywhere

Free · No signup · No trial clock

SEE THE DIRECTORY