메뉴
Read reply emails and classify them by intent (interested, objection, not now, unsubscribe) with recommended next actions.
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
SOC 직업 분류 기준
| name | reply-classifier |
| description | Read reply emails and classify them by intent (interested, objection, not now, unsubscribe) with recommended next actions. |
Read incoming reply emails from outbound campaigns and classify each by intent. Output a categorized list with recommended next actions so your team responds fast and appropriately.
from_email, from_name, subject, body, received_date, campaign_name.eml or .txt files| Category | Description | Response SLA |
|---|---|---|
| Interested | Wants to learn more, asked a question, agreed to meet | Within 1 hour |
| Objection | Raised a concern but didn't shut the door (budget, timing, competitor, authority) | Within 4 hours |
| Not now | Timing isn't right but didn't say never ("circle back later", "not a priority right now") | Acknowledge + set reminder |
| Referral | Pointed you to someone else ("talk to my colleague", "reach out to X") | Within 2 hours |
| Unsubscribe | Wants to be removed ("remove me", "stop emailing", "unsubscribe") | Immediate - remove and confirm |
| Out of office | Auto-reply, OOO message | No action - re-engage after return date |
| Not relevant | Wrong person, wrong company, spam, bounce notification | Archive |
For each classified reply, generate:
Flag edge cases: replies that are ambiguous or could be read multiple ways → mark as "needs human review"
CSV with columns:
from_email, from_name, company, received_date, category, confidence, suggested_response, crm_action, campaign_name
Plus a summary:
Total replies processed: N
- Interested: N (respond within 1 hour)
- Objection: N (respond within 4 hours)
- Not now: N (set reminders)
- Referral: N (follow up on referral)
- Unsubscribe: N (remove immediately)
- Out of office: N (no action)
- Not relevant: N (archive)
- Needs human review: N
Save to: ./data/replies/classified_[DATE].csv
#replies channel within 1 minute for fastest response. Include: sender name, company, reply snippet, suggested response, and CRM action.#replies with suggested rebuttal from copy frameworks.Configure Slack webhook: set SLACK_WEBHOOK_URL in your .env file.
/reply-classifier
Input: ./data/replies/inbox_export_march_10.csv
Output:
Total replies: 24
- Interested: 5 → respond ASAP
- Sarah Kim (DataFlow): "This looks relevant. Can you send a one-pager?"
→ Suggested: "Sure thing, Sarah. Here's our one-pager [LINK]. Happy to walk through it live - does Thursday 2pm work?"
→ CRM: Move to "Engaged", set task for Thursday
- Objection: 3
- Mike Chen (Acme): "We just signed a 2-year deal with [COMPETITOR]."
→ Suggested: "Makes sense - [COMPETITOR] is solid for X. Where we're different is Y. Worth keeping in touch for when the renewal comes up?"
→ CRM: Set "Competitor" field, set reminder 18 months out
- Not now: 6
- Alex Rivera (CloudCo): "Not a priority this quarter."
→ Suggested: "Totally get it. I'll circle back in Q3. In the meantime, here's a relevant case study if useful."
→ CRM: Set follow-up task for July 1
- Unsubscribe: 2 → remove from all sequences immediately
- Out of office: 4 → no action, earliest return: March 17
- Referral: 1 → follow up on referral to VP Engineering
- Not relevant: 2 → archive
- Needs human review: 1 → tone unclear, may be sarcastic
Build a complete micro-campaign from ICP segment definition to campaign-ready output files.
Enrich a list of companies or contacts with firmographic, technographic, and contact data using available APIs.
Generate personalized first lines, email copy, or LinkedIn messages based on prospect data and copy frameworks.
Analyze CRM pipeline data to identify risks, recommend actions, and generate a weekly pipeline summary.
Scan target accounts for new buying signals and output a prioritized alert list.