Safety rules
Limits SendCrew enforces on every sender. No API or MCP parameter can change them.
These rules run on the server for every write, whoever sends it. They are what keeps a fast agent from getting a LinkedIn account restricted.
Caps and warmup
New senders start low and ramp up to the warm maximum over 14 to 21 days (older LinkedIn accounts warm faster). A sender whose health drops is throttled further on its own.
| Per sender, per day | Warmup start | Warm maximum |
|---|---|---|
| Connection requests | 5 | 25 |
| Messages | 15 | 50 |
| InMails | 2 | 10 |
| Profile views | 20 | 80 |
| Reactions and comments | 8 | 30 |
| Invite withdrawals | 10 | 20 |
- Connection requests are also capped at 100 per sender per week.
- Actions run between 8:00 and 19:00 in the sender's timezone. Weekends run at half caps.
- Each sender waits a random 45–180 seconds between actions.
Never twice
Once a sender contacts a person, that person belongs to that sender in your workspace. Another sender trying to invite or message them is refused with SAFETY_BLOCKED. Contacts are never shared between workspaces.
Circuit breakers
If LinkedIn shows a checkpoint, a captcha, a restriction or an expired session, SendCrew:
- pauses the sender,
- leaves its pending actions waiting (nothing retries through a checkpoint),
- sends a
sender.pausedwebhook and emails the workspace owners.
Resolve it in LinkedIn, reconnect the sender in the dashboard, and its queue continues.