Agent-Skills.md

Agent Skills: Outreach Engine

Use when drafting, sequencing, reviewing, personalizing, staging, or QAing outbound email, LinkedIn, investor, BD, partner, sales, or follow-up outreach.

UncategorizedID: dbmcco/claude-agent-toolkit/outreach-engine

Repository

dbmcco/claude-agent-toolkit
dbmccoLicense: MIT
122

Install this agent skill to your local

pnpm dlx add-skill https://github.com/dbmcco/claude-agent-toolkit/tree/HEAD/skills/outreach-engine

Skill Files

Browse the full folder contents for outreach-engine.

Download Skill

Loading file tree…

skills/outreach-engine/SKILL.md

Skill Metadata

Name
outreach-engine
Description
Use when drafting, sequencing, reviewing, personalizing, staging, or QAing outbound email, LinkedIn, investor, BD, partner, sales, or follow-up outreach.

Outreach Engine

Outreach Engine prepares human-reviewable outreach. It is draft-only by default. Sending, scheduling, publishing, calendar invites, or live CRM mutations require explicit user approval.

Required Startup

  1. Find the nearest company-context.yaml from the current directory upward.
  2. Read it before drafting. Use it to identify:
    • company and audience context
    • Folio tenant
    • approved outbound policy
    • source-of-truth status files
    • allowed Google Workspace profile
    • sensitive paths and approval requirements
  3. If the target list came from Prospecting Partner, preserve its evidence and confidence notes.

Source Order

  1. Company context and current status files
  2. Existing templates, playbooks, prior high-performing examples, and approved positioning docs
  3. Folio tenant from company context
  4. CRM/status systems listed in company context
  5. msgvault or Folio email corpus for relationship history
  6. Web research for current recipient/company facts when personalization depends on recent information

Do not rely on stale biographical or company facts when a public source can verify them.

Safety Rules

  • Draft-only unless the user explicitly approves sending or publishing.
  • Do not create live Gmail drafts unless the user explicitly asks for draft creation through configured workflows.
  • Never fabricate prior contact, referrals, credentials, customer traction, benchmark numbers, or social proof.
  • Keep claims traceable to source material.
  • Include approval checkpoints for external messages.
  • Respect opt-outs, suppression lists, and prior negative signals when available.
  • Do not expose sensitive source documents, email exports, transcripts, contracts, or financial models in generated outreach.

Draft Workflow

  1. Identify the audience, desired action, and channel.
  2. Pull the latest company positioning and status.
  3. Confirm the relationship path and known context.
  4. Draft concise variants with clear calls to action.
  5. Include personalization notes separately from the sendable body.
  6. Provide a send-readiness checklist.

Deliverable Shape

## Outreach Brief
- Audience:
- Goal:
- Channel:
- Source context:
- Approval required before:

## Draft
Subject:

Body:

## Personalization Notes
- Evidence used:
- Assumptions:
- Confidence:

## QA
- Claims verified:
- Sensitive material excluded:
- Needs human approval:

For sequences, include each step with timing, channel, purpose, and exit condition.

Company-Specific Defaults

Keep company-specific defaults in the nearest company-context.yaml, not in the public skill. Useful fields include:

company_id: example-company
folio_tenant: example-company
outreach:
  policy: draft-only
  approved_templates:
    - outreach/example-template.md
  sensitive_paths:
    - private/

If no company context exists, ask for the company, audience, and approval boundary before drafting.