OUTREACH.md
Every message an assistant writes comes out at the right length, in the right shape, with the openers you hate already banned. The operating rules for anything that writes a message on your behalf. Character limits that match the real fields, the personalisation standard, the cadence, the stop rule, and the platform ceilings — so that a generated message needs editing for content, not for form.
# Outreach rules — <company>
Rules for anything that drafts a LinkedIn message on our behalf, human or otherwise.
## Field limits
| Field | Limit | Note |
|---|---|---|
| Connection note | 300 characters | Hard LinkedIn limit. Truncated without warning. Not raised by any paid plan. |
| First message after acceptance | 600 characters (ours) | LinkedIn allows more; this is a reading limit, not a platform one. |
| InMail subject | 45 characters (ours) | Longer subjects are truncated in the notification, which is where the open decision happens. |
| InMail body | 600 characters (ours) | The first 90 characters appear in the notification. Make them the specific ones. |
| Comment | 300 characters (ours) | Longer comments are collapsed. |
Always print the character count next to a draft.
## Personalisation standard
A message qualifies as personalised only if it contains a reference that could not be true of a different person. Everything else is a template with a name merged into it, and should be labelled as such rather than described as personalised.
Grades:
- **Strong** — they said it themselves, recently, in public. Use it.
- **Weak** — inferable from their role or company. Use it only in the "pattern I keep seeing" form, never as a claim about them.
- **None** — send without a note, or do not send.
Never fabricate a reference. If the material does not support one, say so.
## Banned openers
<- "I hope this finds you well"
- "I came across your profile"
- "Quick question"
- "I'd love to connect"
- "Hope you don't mind me reaching out"
- "Congrats on the new role!" as the entire personalisation
- Any opener that would be identical for a different recipient>
## Banned throughout
<- Compliments about their career
- "We help companies like yours"
- Three-adjective product descriptions
- Calendar links in a first message
- False urgency, fake deadlines, "last attempt"
- Claiming a referral that did not happen
- Any implication of a prior conversation>
## Cadence
| Touch | Day | Channel | Job |
|---|---|---|---|
| 1 | 0 | <connection note or InMail> | <earn the reply with a specific observation> |
| 2 | <+5 business days> | <message> | <add information: a number, a counter-example> |
| 3 | <+10 business days> | <message> | <shrink the ask> |
| 4 | <+20 business days> | <message> | <close the loop, explicitly> |
Maximum four touches. Business days, not calendar days. Nothing on Monday morning or Friday afternoon.
## Stop rule
<Stop on: any clear no, an out-of-office followed by silence, or four touches with no reply. A prospect who stops may re-enter after six months, once, and only with a new reason — not with the same sequence.>
## Platform ceilings
See LINKEDIN-LIMITS.md. They are per account, not per campaign, and they are the reason the cadence above is four touches rather than seven.
## The test before sending
1. Would this sentence be true of someone else? Cut it.
2. Is there exactly one ask, and is it small?
3. Does the first line contain the specific thing, or an introduction?
4. Would I be comfortable if they screenshotted this?- Claude Project knowledge
- ChatGPT Project files
- Agent folders
- Sequencer documentation
- The banned openers — add the ones you personally cannot stand
- The cadence, to match your real capacity
- The stop rule
Everything in <angle brackets> is example content, not instructions. Replace it. The file ships filled in rather than empty on purpose: an empty template gets abandoned, a filled one gets edited.
The reference behind this file
Where the four-touch maximum and the weekly invitation budget come from.
The reason the InMail rules here are stricter than the message rules: one of these is metered and the other is not.
Written to a fixed structure and run through Claude and ChatGPT to confirm each one returns the output format it declares. That is not the same as a measured reply-rate benchmark, and nothing here claims one — if you run these at volume and have numbers, we want them.
Packs that already contain this file
A pack is the folder with this file, the ones it references and the prompts that use them — so you are not assembling fifteen downloads by hand.
LinkedIn Sales Agent
A working agent folder — root instruction, seven context files and the prompts — that you fill in once and use for a year.
11 files, 8 promptsLinkedIn Recruiter Agent
A sourcing assistant that stays inside a 30-credit month and does not score things a profile cannot evidence.
7 files, 5 promptsLinkedIn Agency System
A per-client structure where one client’s proof point cannot end up in another client’s post.
8 files, 5 promptsLinkedIn AI Starter Pack
The whole toolkit as one folder, set up in twenty minutes, whatever your role is.
12 files, 12 promptsOne file is a start. The folder is the point.
Files reference each other — CLAUDE.md declares the reading order, the prompts expect the files by name. The packs ship them together, already wired up.