course module

Bucket 3 — Two Memory Prompts (Use /remember and /recall)

Prompts 6 & 7 — Voice Lock and Personal Index: the two prompts that turn the agent into something a chat tab can't be. Uses MAKO's /remember and /recall, built in.

ai-tools free

MAKO ships with /remember <name> <content> and /recall <query> out of the box — it's a persistent Postgres-backed knowledge base. These two prompts use it to build memory that compounds.

Prompt 6 — Voice Lock

Send this once, then /remember voice-profile <the agent's answer>:

I'm going to paste 10 things I've written publicly — tweets,
LinkedIn posts, emails. Build a voice profile from them: my
cadence, recurring phrases, what I never say, the rhythm of my
sentences, who I'm usually talking to.

[paste 10 things here]

The agent gives back a profile. You save it:

/remember voice-profile <paste the profile back>

From now on, every time you ask the agent to draft something, prefix it with: "Use the voice-profile from /recall voice to draft this."

This is the one prompt that turns the agent from "writes generically" to "writes like you". Drafts stop needing rewrites.

Prompt 7 — Personal Index

Same pattern. Send this once:

I'm going to give you the things I want you to remember about my
world: my top 10 high-trust people (who they are and how I know
them), my 5 active deals, my 3 main projects, my offer stack,
and my rough financial runway.

[paste all of that]

Compress this into a structured index I can save and retrieve.

Then save:

/remember personal-index <paste the structured index>

Now reference it freely: "Pull /recall personal-index and tell me which of my deals I haven't touched in 7 days based on our conversation history."

The agent recalls the context proactively. You stop briefing it every time. Week 1 you build the index. Week 4 you get back insights you didn't ask for because the agent connected dots across conversations.

Wrap-up — What To Do Tonight

The seven prompts above are the foundation. They earn their keep in week one, all out of the box.

Pick one — just one. The highest-leverage starter is Prompt 6 (Voice Lock) — it makes every future prompt better. Then Prompt 2 (Weekly Review) on Friday.
Run it. Open MAKO on Telegram, send the prompt, save the output with /remember if it's a memory prompt.
Refine in week 2. Once you see what sticks, write your own prompt 8. The pattern is clear by then.

If you don't have an agent running yet — HireMako gets you live on Telegram in 15 minutes — Claude Opus under the hood, /schedule, /remember, voice messages, all on your own infra. Built to run Drop-in Prompts from Day 1.

Next up

Bucket 4 — Tools & Integrations (When Your Agent Touches Real Systems)

Same agent, same prompt — one has tools. The four tool categories (retrieve, write-back, notify, act), the guardrail ladder, one live integration, and the three-question test before you wire anything in.

I'm going to paste 10 things I've written publicly — tweets, LinkedIn posts, emails. Build a voice profile from them: my cadence, recurring phrases, what I never say, the rhythm of my sentences, who I'm usually talking to. [paste 10 things here]

I'm going to give you the things I want you to remember about my world: my top 10 high-trust people (who they are and how I know them), my 5 active deals, my 3 main projects, my offer stack, and my rough financial runway. [paste all of that] Compress this into a structured index I can save and retrieve.