reply-classification

Name: reply-classification
Author: Xuxchloris

17stars2forks

How to install reply-classification?

xuxchloris/export_skills/reply-classification

$npx -y skills add Xuxchloris/export_skills --skill reply-classification

Installs into the current project.

Summary

Use when a trade agent needs to classify buyer replies, inquiry responses, catalog requests, sample requests, quotation requests, objections, unsubscribe messages, or unclear inbound email

Files · 1

View on GitHub

SKILL.md

1	# Reply Classification
2
3	## Overview
4
5	Classify buyer replies into a clear sales status and recommend the next human-reviewed action. The core rule is to preserve the buyer's intent without inventing commitment or urgency.
6
7	## When to Use
8
9	Use this skill after an inbound buyer email, message, or reply arrives and the user wants to know what it means or what to do next.
10
11	Use `email-crafting` after classification when a reply draft is needed.
12
13	## Inputs
14
15	- Buyer reply text
16	- Prospect score and company research output
17	- Previous outreach summary
18	- Product and pricing context, when relevant
19
20	## Outputs
21
22	```json
23	{
24	"classification": "inquiry\|quotation_request\|catalog_request\|sample_request\|meeting_request\|objection\|not_interested\|unsubscribe\|out_of_office\|unclear",
25	"confidence": "low\|medium\|high",
26	"buyer_signals": [],
27	"requested_items": [],
28	"missing_information": [],
29	"recommended_action": "",
30	"reply_draft_needed": false,
31	"human_review_required": true
32	}
33	```
34
35	## Procedure
36
37	1. Extract the buyer's explicit request, product references, quantities, destination, deadline, and questions.
38	2. Select one classification that best matches the reply. Use `unclear` when the message needs human judgment.
39	3. Record buyer signals as observed facts only.
40	4. List commercial details that still need confirmation.
41	5. Recommend one next action: reply, prepare quotation draft, send catalog draft, discuss sample, schedule meeting, pause follow-up, or request human review.
42	6. Send quotation requests to `quotation-generator`.
43	7. Send reply drafting requests to `email-crafting`.
44
45	## Verification
46
47	- Classification matches the buyer's explicit message.
48	- Buyer intent is not overstated.
49	- Missing quantity, SKU, destination, or commercial terms are listed when relevant.
50	- Recommended action is specific and human-reviewed.
51	- Unclear replies remain `unclear`.
52
53	## Common Mistakes
54
55	\| Mistake \| Fix \|
56	\| --- \| --- \|
57	\| Treating a polite reply as buying intent \| Record the exact buyer signal \|
58	\| Drafting a quote without required details \| Pass missing fields to quotation workflow \|
59	\| Guessing the requested product \| Ask for clarification \|
60	\| Continuing outreach after unsubscribe \| Recommend pausing follow-up \|

People who liked this also viewed

6 picks

Type

reply-classification

17stars2forks

View on GitHub

How to install reply-classification?

xuxchloris/export_skills/reply-classification

$npx -y skills add Xuxchloris/export_skills --skill reply-classification

Installs into the current project.

Summary

Use when a trade agent needs to classify buyer replies, inquiry responses, catalog requests, sample requests, quotation requests, objections, unsubscribe messages, or unclear inbound email

Files · 1

View on GitHub

SKILL.md

1	# Reply Classification
2
3	## Overview
4
5	Classify buyer replies into a clear sales status and recommend the next human-reviewed action. The core rule is to preserve the buyer's intent without inventing commitment or urgency.
6
7	## When to Use
8
9	Use this skill after an inbound buyer email, message, or reply arrives and the user wants to know what it means or what to do next.
10
11	Use `email-crafting` after classification when a reply draft is needed.
12
13	## Inputs
14
15	- Buyer reply text
16	- Prospect score and company research output
17	- Previous outreach summary
18	- Product and pricing context, when relevant
19
20	## Outputs
21
22	```json
23	{
24	"classification": "inquiry\|quotation_request\|catalog_request\|sample_request\|meeting_request\|objection\|not_interested\|unsubscribe\|out_of_office\|unclear",
25	"confidence": "low\|medium\|high",
26	"buyer_signals": [],
27	"requested_items": [],
28	"missing_information": [],
29	"recommended_action": "",
30	"reply_draft_needed": false,
31	"human_review_required": true
32	}
33	```
34
35	## Procedure
36
37	1. Extract the buyer's explicit request, product references, quantities, destination, deadline, and questions.
38	2. Select one classification that best matches the reply. Use `unclear` when the message needs human judgment.
39	3. Record buyer signals as observed facts only.
40	4. List commercial details that still need confirmation.
41	5. Recommend one next action: reply, prepare quotation draft, send catalog draft, discuss sample, schedule meeting, pause follow-up, or request human review.
42	6. Send quotation requests to `quotation-generator`.
43	7. Send reply drafting requests to `email-crafting`.
44
45	## Verification
46
47	- Classification matches the buyer's explicit message.
48	- Buyer intent is not overstated.
49	- Missing quantity, SKU, destination, or commercial terms are listed when relevant.
50	- Recommended action is specific and human-reviewed.
51	- Unclear replies remain `unclear`.
52
53	## Common Mistakes
54
55	\| Mistake \| Fix \|
56	\| --- \| --- \|
57	\| Treating a polite reply as buying intent \| Record the exact buyer signal \|
58	\| Drafting a quote without required details \| Pass missing fields to quotation workflow \|
59	\| Guessing the requested product \| Ask for clarification \|
60	\| Continuing outreach after unsubscribe \| Recommend pausing follow-up \|

People who liked this also viewed

6 picks

Type