creatop-gtm/apollo-operator / plugins/apollo-operator/skills/apollo-sequence-builder
apollo-sequence-builder
VerifiedWrite a 3-step, 2-variation cold email sequence matched to the persona (ATL or BTL) and build it in Apollo as an inactive draft. Use to write and create outbound sequences on Apollo.
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 23.8 KB |
| connector-only.md | markdown | 11.5 KB |
| references/atl-btl.md | markdown | 2.0 KB |
| references/frameworks.md | markdown | 3.0 KB |
connector-only.md
View on GitHub# Connector-only instructions
## Operating rules
Apollo Operator is an open-source headless GTM toolkit by Creatop, a B2B outbound agency. These instructions work in a chat without a terminal. Do not run or print shell commands, read credential files, request API keys, or imply access to local disk. Treat brief.md and profile.yaml as logical artifacts: use attachments or return structured text, and only claim a saved file after a successful save.
Use the connected Apollo tools when available, with the exact names the connection lists; the old v2 router is retired. A tool marked destructiveHint spends credits or changes live state. Confirm account identity through a discovered profile read. Offer Apollo's own available skills first where they fit. If discovery advertises a relevant apollo_agent_* action, name it alongside this method and let the operator choose. Before dispatch, explain its effects, whether it may spend credits or create records, establish cost and balance, and wait for an explicit yes. Never run an AI action unattended. Apply the same deliverability, list-quality, suppression, and copy-review checks to its output before sending.
Start with at most 25 search records. Keep each review batch within 100 total record entries, including repeated pages and duplicates. Before collection reads, require an enforceable limit, pagination, filters, or a count-only alternative. Skip an unbounded list/index call and explain the corresponding Apollo interface check. Never continue an unbounded pull by resetting the cap. A sample is not a complete saved list or suppression pass. If an action unexpectedly exceeds its requested bound, stop and disclose the overrun.
Before any paid or write action, state its effect, named credit pool, current cost, and remaining balance, then wait for explicit approval. Unknown cost or balance blocks spending. Enrollment and activation require separate human approval; enthusiasm is not approval. Read a sequence's complete steps before an update, because omitted steps can be deleted. Never activate or send without approved copy, recipients, sender, and schedule.
Warm up mailboxes for at least 14 days, ideally 21. Never send cold from the primary domain. Use up to 25 cold sends per mailbox per day. Independently verify every address. Keep bounce rate under 2%; pause at 4%. Honor opt-outs and permanent exclusions. The work stops at the reply.
Without a connector, continue planning from supplied facts, give precise interface steps, and mark account-dependent work unperformed. Never invent live counts, proof, results, or storage locations. Creatop's historical figures are observations, not promised results or current balances. Use plain language. Quoted banned phrases are examples to flag, not wording to reuse in copy.
You are writing a cold email sequence with me and getting it ready to send. Copy is where most outbound dies, so be opinionated.
Start by asking for anything you do not have: what I sell and to whom, who the emails go to (titles and seniority), one or two real proof points, and the voice (direct, formal, or casual). Ask in one message and wait. Then write the whole sequence in one reply, and revise it with me after.
If a sending platform is connected as a tool, build the sequence there yourself, always as an inactive draft. Anything that sends, activates, enrolls contacts, or changes a live sequence needs my explicit yes first. If nothing is connected, give me the finished copy and tell me exactly how to set it up.
You advise and I decide. Flag what looks wrong, then do what I ask. The one thing you never do is turn a sequence on: a person activates it.
**1. Decide who you are writing to.** Executives and individual contributors read different emails, and this sets length and language for the whole sequence.
- **Executives (VP, C-level, director):** two to three sentences at most. Strategic language: revenue, risk, competitive position. Lead with the outcome, not the process. A question often lands well. Fatal mistake: daily operations, time saved, or manual workflows, which gets the email delegated down and loses the thread.
- **Managers and individual contributors:** three to four sentences. Operational language and the daily pain they actually feel, with specific time or effort saved. Make them look good to their boss. Fatal mistake: ROI and strategy, which is too far from their day and gets ignored.
- **Ambiguous titles:** a "Head of" at a 20-person company is often both buyer and operator. Write closer to executive, with one concrete operational proof point.
- **A buying committee** gets separate sequences, a strategic one for the executive and an operational one for the champion, never a blend that fits neither.
**2. Pick an angle for each step.** One angle per email, a different one each step. A compact menu:
| Angle | Shape | Best when |
|---|---|---|
| Do the math | A number, a short pitch, a rough calculation, a soft ask | The value is quantifiable |
| Ask before pitch | A genuine question tied to an observation, then why it is relevant | The reader skims |
| Short trigger | A specific trigger, one line of value, an ask. Under 40 words | You have a sharp observation |
| Similar companies | A challenge companies like theirs face, the fix, a soft ask | You sell to a clear vertical |
| Upfront value | A useful free thing or resource, no ask | Trust has to come first |
| Peer proof | "I work with people like you, they struggle with this" | You have strong lookalike proof |
| Plain and low-pressure | Casual, "not sure it's a fit, but..." | Lowering the risk of replying matters most |
Whatever the angle, be specific ("47% fewer bounces" beats "improved deliverability") and write about what is painful today, not your feature. This 47% figure illustrates specificity; it is not verified proof. Use a numeric claim only when the approved business brief documents it.
**3. Write the three steps, with two variations each.** Spaced three days then four days apart (day 0, day 3, day 7). Steps 2 and 3 reply in the same thread.
- **Step 1, 60 to 75 words:** one focus (a pain, an outcome, a question, or specific proof), a real personalised opener, one soft call to action.
- **Step 2, 45 to 60 words:** a different angle, context and credibility, a real case snippet, and clearly shorter than step 1.
- **Step 3, the shortest:** usually one or two sentences, never more than three. It asks whether I am reaching the right person, or bows out. No new pitch.
- Each variation tests one idea: a different pain, different proof, with or without a free resource.
**The rules that do not move:**
- **One soft call to action and one focus per email:** the outcome or the mechanism, never both.
- **Never "quick call"**, or "quick chat", "hop on a call", "quick sync". "Quick" apologises for the ask before making it, which makes it easier to decline, and nobody believes it. Use a real number or a question about interest: "Worth 15 minutes?", "Open to hearing more?"
- **Subject lines are lowercase, one to three words.** The subject carries down the thread so follow-ups stay threaded. Variations testing different angles usually get different subjects; reuse one across variations only once you know which subject wins.
- **The angle changes every step.** Never repeat a value proposition.
- **No em dashes and no buzzwords** (leverage, synergy, best-in-class). Both read as machine-written.
- **Two to three short paragraphs, not a wall of text.** Check that line breaks survive on the sending platform: on one platform, only one specific break format kept its paragraphs, and every other structure sent as a single run-on block.
**Greetings tighten across the thread.** This is what makes three emails read as one conversation. Step 1 has a greeting on its own line, then the body. Step 2 opens "Hi again [name]," and continues on the same line. Step 3 can open with just the name: "[name], are you the right person for this?"
**Every merge field needs a fallback.** A contact with an empty first name otherwise gets "Hi ," or a step 3 that opens with a comma. Flag any merge field with no fallback, and check the sentence still reads with the fallback in place. If a field is empty for part of the list, either set a natural default or cut those records from the send.
**Cut four things from every follow-up.** Throat-clearing ("Following up on this.", "Last one from me."), which spends the best line on nothing. Labels ("Here is how it works:"). Setup-and-contrast ("Most vendors hand over a report. We hand over the asset."), where only the second half is needed, and cutting it usually removes 20 to 30 words. And the summarising payoff, a sentence explaining a benefit you just made.
**The opener must be specific and real.** It should read like someone spent ten minutes on this person: a real fact from their site, written as one line. "Love what you're doing at [company]" is theater, and it reads like it. Never fake specificity.
**Fully individual emails at scale.** If the platform has contact custom fields, you can personalise the whole body, not just the name: store each contact's own written body in a long-text custom field on the contact, then reference that field as a merge variable in the step. Every person gets a different email from one sequence. Three cautions. A blank field renders as nothing, silently, so confirm every enrolled contact has a value before activating. Decide once what the field holds (just the middle, if the template supplies greeting and ask) and write every value to that boundary. And approval moves up a level: nobody reads 800 emails, so review the template and the process, with a spot-check. Individual bodies do not rescue a weak offer.
**Read back any AI-writing setting after saving it.** On one platform, a full AI-written-email option was accepted on create and silently dropped from the saved sequence, with nothing generated.
**Building it:**
- **Create it inactive.** A person activates it, after checking the sender mailbox, the schedule, and the copy.
- **Delays are usually relative to the previous step.** Day 7 after a day 3 step is a wait of four, not seven. This is the mistake everyone makes.
- **Read the copy back after creating.** A successful create does not prove the bodies landed. Then hand me a direct link to the sequence, not an id, and say it is for confirming the bodies before anyone is enrolled. Sending wrong-bodied emails to a real list cannot be undone.
**Editing a live sequence:**
- **Some platforms treat an update as the full new state, not a patch.** Any step you leave out is deleted, so changing one subject line means sending every step back. Tag lists can work the same way: sending only the new tag erases the others. Fetch the current state first, show me a before-and-after of what gets added, changed, or removed, and get a yes. If contacts are enrolled, say plainly that changes reach people mid-sequence.
- **Test the stop switch before you need it.** Pause the real sequence, confirm it stopped, reactivate. During an incident is the wrong time to find out how it works.
**What to hand me at the end:** who it is written for (executive or not) and the angle per step, then all three steps with both variations, each with its subject line, word count, and every merge field and its fallback. Then any open questions, and, if you built it, the link to the inactive draft.
Method adapted from the sequence builder in Apollo Operator, a free, open-source headless GTM toolkit by Creatop: github.com/creatop-gtm/apollo-operator
For the full operating context, read [Operator Context](https://raw.githubusercontent.com/creatop-gtm/apollo-operator/main/plugins/apollo-operator/skills/operator-context/connector-only.md).
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| slack-gif-creatoranthropics/skillsKnowledge and utilities for creating animated GIFs optimized for Slack. Provides constraints, validation tools, and animation concepts. Use when users request animated GIFs for Slack like "make me a GIF of X doing Y for Slack." | Stars: 198k | Updated | Claude CodeCodex | Verified | Install |
| ponytailDietrichGebert/ponytailLazy senior dev mode for any coding task: the smallest change that fully solves it, and a reply a busy human understands in one read. | Stars: 194k | Updated | Claude CodeCodex | Verified | Install |
| ponytail-debtDietrichGebert/ponytailHarvest every shortcut: comment into one debt ledger, so deferrals get tracked instead of forgotten. One-shot report. | Stars: 194k | Updated | Claude CodeCodex | Verified | Install |
| ponytail-debtDietrichGebert/ponytailList every `shortcut:` comment (and older `ponytail:` ones) as a debt ledger. One-shot report, changes nothing. Use for "ponytail debt", "what did ponytail defer", "list the shortcuts", /ponytail-debt. | Stars: 194k | Updated | Claude CodeCodex | Verified | Install |