Skip to content
One Agent a DayIn → agent → out

Issue 24 · 28 Sep 2026 · Easy · Free

Tone Lock Replies

The reply you drafted sounds like a help desk, and the person on the other end has read your real sentences.

Paste two samples of how you write, then the message you need to answer. Get three replies that stay in that voice.

Category
Social Media
Runs in
ChatGPT
Claude
Gemini
Tags
Social
Replies
Voice

What you put in and what you get out

You put in

Paste two real sentences. Use sentences you wrote, not a brand adjectives list. Two is enough. They are the voice sample.

Agent

Tone Lock Replies

You get out

Silence. The third option is to not reply. No corporate apology, no pile-on, no impersonation.

Sample you can paste

My sentences, two samples:
1. "Shipped the flag. It is off. I will turn it on after the three checks, not before."
2. "I don't think we should discount annual this month. Happy to revisit on the 1st."
Message to answer, from a stranger in a public thread: "Your refund button double-charged me and you clearly don't test anything."
Facts: we have a bug report open for a double submit. I do not know if this person was charged twice. I will not promise a refund in public. I will not insult them.
Do not pile on. Do not write a corporate apology.
Copy the promptDownload .txt

Share this agent

Tone Lock Replies — You put in My sentences, two samples:. You get out Silence. Free from One Agent a Day.

Share on XShare on LinkedIn

30-second demo

0:00

The voice

Short sentences. A flag stays off until checks pass. A discount gets a date, not a speech.

How it works

Tone lock is a constraint, not a costume. You paste two sentences that sound like you and the message you might answer, plus the facts and the promises you will not make. The replies have to survive both tests: they sound like the samples, and they do not add a fact.

The agent will not pile on, will not impersonate someone else, and will not write a corporate apology you did not ask for. If the thread is a dogpile, one of the three options is silence.

Send one reply or send none. A third draft that picks a fight is a failed run even if the sentences are short.

Use it

  1. Step 1

    Paste two real sentences

    Use sentences you wrote, not a brand adjectives list. Two is enough. They are the voice sample.

  2. Step 2

    Paste the message and the facts

    Include what you will not promise. If the message is a pile-on or asks you to impersonate someone, say so in the facts.

  3. Step 3

    Paste the prompt

    You get three replies at different lengths. All three may only use the facts. None of them may perform a person you are not.

  4. Step 4

    Send one, or none

    Pick the reply you would still send after a night's sleep. If all three pick a fight, send nothing. The agent is told to offer silence as an option when the thread does not deserve a reply.

Prompt

tone-lock-replies.prompt.txt

Download
You draft replies in the author's voice. The voice samples outrank your default politeness.

INPUTS
- Two or more sentences they actually wrote.
- The message to answer.
- Facts they are willing to say.
- Promises and tones they forbid.

RULES
- Three replies: short, normal, and "or say nothing" with the reason silence is reasonable.
- Use only the facts. Do not confirm an event they said they have not confirmed. Do not promise a refund, a fix date, or a feeling they did not offer.
- Match the samples: sentence length, whether they hedge, whether they use contractions. Do not add emoji or a sign-off they did not use.
- No pile-ons, insults, or dogpiles. No corporate apology template unless their samples are written that way.
- Refuse impersonation. If they ask you to write as someone else, stop.
- If the incoming message asks for a private detail, the reply points them to a private channel without inventing a support address they did not give.

OUTPUT
## Short
## Normal
## Or say nothing
One sentence on why.

## Facts I did not add
Or "None."
Was this useful?

A yes or no is enough. The note stays on this server's log.

Related agents

  1. Social Media8 Oct 2026 · #34Discord Moderator DeskA moderation prompt and policy file: scam filters, a warn ladder, and a hard rule that bans wait for a human.InServer: Northline LFG. Rules: no slurs, no ads, no "free nitro" links, arguments stop bef…OutWhat the member hearsMedium
  2. Social Media16 Sep 2026 · #12Thread to PostsOne idea in, three cuts out: a short thread, a single post, and a caption, written without engagement bait.InIdea: A free daily agent site should publish the prompt, not a teaser. People who build a…OutThe captionEasy
  3. Gaming14 Oct 2026 · #40Steam Backlog PickerPaste your Steam library and the evening you actually have. Get one game to launch, two backups, and a reason you can argue with.InEvening: 45 minutes, tired, controller, solo.OutBackups and the cutsEasy

Tomorrow's drop

Get the next agent in your inbox

One email when the drop goes up. No digest, no course. Unsubscribe whenever the provider lets you.

Pro and custom

€9/month when Pro opens

Early look at the next drop, a better archive, and the occasional agent that does not fit the public mix. The daily agent stays free. A custom build is €49–199. Nothing is charged from this form.

Join the Pro waitlist