一键导入
build-contact-list
Build a target contact list and run contact-level signals against them. Works with the Saber CLI, Apollo, LinkedIn, HubSpot, or manual research.
用 Codex 或 Claude 帮你安装 复制这段 Prompt,粘贴到 Codex、Claude 或其他助手里,让它检查 Skill 页面并帮你完成安装。
菜单
Build a target contact list and run contact-level signals against them. Works with the Saber CLI, Apollo, LinkedIn, HubSpot, or manual research.
用 Codex 或 Claude 帮你安装 复制这段 Prompt,粘贴到 Codex、Claude 或其他助手里,让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Activate company-level signal tracking. Runs signals via the Saber CLI if available; without it, outputs a research playbook and suggests alternative tracking methods.
Generate 12–15 structured, weighted research signals from a structured ICP, ready to activate in Saber. Each signal includes answer type, weight, and interpretation rules for automatic scoring.
Deep research on a single company — signals, hiring, news, tech stack, and LinkedIn presence. Uses the Saber CLI for signal data and any available MCP tools for broader research.
End-to-end guided workflow — take one prospect domain, run your saved research signals, find named contacts on the buying committee, and write a hyper-personalised cold email and LinkedIn DM grounded in the signal answers. First run defines the source-company ICP, contact-search targets, and ≤5 weighted signals (saved to disk); every subsequent run reuses them. Use when you want full SDR research-and-write for a single prospect, not write-outreach in isolation.
End-to-end guided workflow — build a Saber company list, run research signals across every company in parallel, find named contacts, and export a combined CSV to the desktop. Walks the user through each stage with confirmations and a credit-discipline guard. Use when you want one signed-off CSV instead of running build-account-list, create-company-signals, and build-contact-list separately.
Build a target account list and run company signals against it. Works with the Saber CLI, Apollo, HubSpot, or any available prospecting tool — with a manual fallback.
| name | build-contact-list |
| description | Build a target contact list and run contact-level signals against them. Works with the Saber CLI, Apollo, LinkedIn, HubSpot, or manual research. |
Use this skill to build a list of target contacts and optionally run signals against them. The Saber CLI is the preferred path, but the skill works with Apollo, LinkedIn, HubSpot, or manual research if you don't have it.
Summarise the contact criteria from conversation context (ICP, persona-research, or ask):
Ask the user to confirm or adjust before proceeding.
Check what's available and take the best path:
Run saber --help to confirm the CLI is installed.
If a target account list exists, get company LinkedIn URLs from it first:
saber list company companies <listId>
Then create the contact list:
saber list contact create --name "<list name>" \
--company-linkedin <linkedin-url> \
--company-linkedin <linkedin-url> \
--title "<title>"
--company-linkedin is required. --title is repeatable for multiple titles.
Use Apollo to search contacts with matching criteria:
If Saber CLI is also available, the contact list can be imported into Saber via the dashboard for signal activation.
Pull contacts from HubSpot matching the persona:
Search HubSpot contacts where jobtitle contains [title], company.numberofemployees between [Y] and [Z]
Work with the results directly. If Saber CLI is available, use LinkedIn URLs from HubSpot to import into Saber.
If no tools are available:
| Name | Title | Company | LinkedIn URL | Email (if known) |
|------|-------|---------|--------------|-----------------|
| ... | | | | |
Suggest using enrich-contacts to add email and phone once the list is built.
Show the list summary and ask the user to confirm before activating signals.
Before running signals across the full list, test 3–5 contacts the user already knows to validate signal accuracy without spending credits on the full list.
For each contact:
saber signal --profile <linkedin-url> --question "<question>" --answer-type boolean --no-wait
saber signal get <signalId>
Present results and ask: do the signals look accurate? Any false positives? Proceed?
If approved contact signals are available in conversation context, offer to run them using create-contact-signals.
After signals complete, present results ranked by intent so the user can sequence outreach.
saber list contact create --name "<name>" --company-linkedin <url> [--title] [--keyword] [--country]
saber list contact get <listId>
saber list contact list