prospect-discovery

Name: prospect-discovery
Author: Xuxchloris

17stars2forks

How to install prospect-discovery?

xuxchloris/export_skills/prospect-discovery

$npx -y skills add Xuxchloris/export_skills --skill prospect-discovery

Installs into the current project.

Summary

Use when a trade agent needs compliant prospect discovery, customer sourcing strategy, lead source planning, search keywords, public directory review, or collection API setup before company research

Files · 1

View on GitHub

SKILL.md

1	# Prospect Discovery
2
3	## Overview
4
5	Create a customer discovery strategy before research and scoring. The core rule is to use approved business sources, record where each lead came from, and pass candidates to `company-research` and `prospect-scoring` for evidence review.
6
7	## When to Use
8
9	Use this skill when the user asks how to find overseas customers, build a prospect list, plan search keywords, use a collection API, review public directories, or prepare leads for outreach.
10
11	Use it for company-level discovery planning and evidence collection before outreach drafting.
12
13	## Inputs
14
15	- Product context from `product-loader`
16	- Market rules from `MARKET.yaml`
17	- Discovery settings from `DISCOVERY.yaml`
18	- Target countries or regions
19	- Approved collection API details, if provided by the user
20
21	## Outputs
22
23	```json
24	{
25	"collection_api_status": "configured\|none\|needs_user_input",
26	"keyword_strategy": [],
27	"target_regions": [],
28	"exclude_terms": [],
29	"allowed_sources": [],
30	"candidate_fields": [],
31	"source_status": "verified\|search_tasks_only\|source_unavailable",
32	"source_evidence_rules": [],
33	"risk_notes": [],
34	"next_steps": ["company-research", "prospect-scoring"]
35	}
36	```
37
38	## Procedure
39
40	1. Read `DISCOVERY.yaml` before asking about collection tooling.
41	2. If `collection_api.provider` is missing, ask the user once for the collection API. If there is no API, write `none` into `DISCOVERY.yaml` when file editing is available.
42	3. If the user provides an approved API after setup, write its provider, endpoint, method, API key environment variable, auth header, query parameters, response mapping, pagination, rate limit, and retry policy into `DISCOVERY.yaml`; keep raw secrets in environment variables.
43	4. Read `scraping.engine`. Use `scrapling-fetcher` only for crawling company pages after a candidate is already known. Do not use browser navigation as the default customer-discovery path.
44	5. Run `python tools/collect_prospects.py --discovery <DISCOVERY.yaml> --product <PRODUCT.yaml> --output-dir <folder>` when file output is needed.
45	6. If the user only names a product or SKU, pass `--product-query` or `--sku` instead of rewriting `PRODUCT.yaml`.
46	7. If the agent has already identified source pages during the task, pass them at runtime with `--source-url` instead of editing `DISCOVERY.yaml`.
47	8. If `discovery_mode` is `native_scrapling_spider`, `scrapling_spider.enabled` is true, or runtime `--source-url` values are provided, `tools/collect_prospects.py` routes to `tools/scrapling_spider_runner.py`, uses Scrapling's native `Spider`, and writes `crawl_report.json`.
48	9. If the agent has MCP access configured, call `tools/scrapling_mcp_server.py` through the `collect_prospects` MCP tool and pass runtime `source_urls` for the same native Spider workflow.
49	10. When neither a configured API, configured `scrapling_spider.source_urls`, nor runtime source URLs are available, create search tasks for Google search results, trade show websites, industry directories, prospect company websites, LinkedIn public summaries, and B2B platform public pages.
50	11. If the tool outputs only `prospect_search_tasks.csv`, stop and report `source_status: "search_tasks_only"`; ask for a collection API, a user-provided list, or permission to process a specific approved source.
51	12. Do not manually browse arbitrary B2B platforms after search tasks are generated.
52	13. Do not return a numbered customer list unless it comes from `prospects.raw.csv`, a user-provided prospect file, or fetched company pages with source URLs.
53	14. If sources cannot be accessed or provide no usable company records, return `source_status: "source_unavailable"` and do not fill the gap with industry knowledge.
54	15. With a configured API, produce `prospects.raw.csv` and `prospects.raw.json`.
55	16. Build keyword groups from product names, HS codes, applications, buyer types, target regions, and channel terms.
56	17. Add exclude terms for jobs, consumer reviews, unrelated retail-only pages, marketplaces without seller websites, and irrelevant industries.
57	18. Keep each candidate tied to a source URL, company-level signal, and reviewable evidence summary.
58	19. When discovery finds a company name or company link, visit the official website or source-linked company page next and run official website contact search before finalizing the row.
59	20. During official website contact search, check homepage, contact, about, team, catalog, and product pages for visible email and phone values.
60	21. For each candidate, record company name, website, country, business type, source URL, evidence summary, risk notes, contact email, contact phone, email result, and phone result.
61	22. If email or phone is not found during discovery, write `contact_email: "没有"`, `contact_phone: "没有"`,

People who liked this also viewed

6 picks

Type

prospect-discovery

17stars2forks

View on GitHub

How to install prospect-discovery?

xuxchloris/export_skills/prospect-discovery

$npx -y skills add Xuxchloris/export_skills --skill prospect-discovery

Installs into the current project.

Summary

Use when a trade agent needs compliant prospect discovery, customer sourcing strategy, lead source planning, search keywords, public directory review, or collection API setup before company research

Files · 1

View on GitHub

SKILL.md

1	# Prospect Discovery
2
3	## Overview
4
5	Create a customer discovery strategy before research and scoring. The core rule is to use approved business sources, record where each lead came from, and pass candidates to `company-research` and `prospect-scoring` for evidence review.
6
7	## When to Use
8
9	Use this skill when the user asks how to find overseas customers, build a prospect list, plan search keywords, use a collection API, review public directories, or prepare leads for outreach.
10
11	Use it for company-level discovery planning and evidence collection before outreach drafting.
12
13	## Inputs
14
15	- Product context from `product-loader`
16	- Market rules from `MARKET.yaml`
17	- Discovery settings from `DISCOVERY.yaml`
18	- Target countries or regions
19	- Approved collection API details, if provided by the user
20
21	## Outputs
22
23	```json
24	{
25	"collection_api_status": "configured\|none\|needs_user_input",
26	"keyword_strategy": [],
27	"target_regions": [],
28	"exclude_terms": [],
29	"allowed_sources": [],
30	"candidate_fields": [],
31	"source_status": "verified\|search_tasks_only\|source_unavailable",
32	"source_evidence_rules": [],
33	"risk_notes": [],
34	"next_steps": ["company-research", "prospect-scoring"]
35	}
36	```
37
38	## Procedure
39
40	1. Read `DISCOVERY.yaml` before asking about collection tooling.
41	2. If `collection_api.provider` is missing, ask the user once for the collection API. If there is no API, write `none` into `DISCOVERY.yaml` when file editing is available.
42	3. If the user provides an approved API after setup, write its provider, endpoint, method, API key environment variable, auth header, query parameters, response mapping, pagination, rate limit, and retry policy into `DISCOVERY.yaml`; keep raw secrets in environment variables.
43	4. Read `scraping.engine`. Use `scrapling-fetcher` only for crawling company pages after a candidate is already known. Do not use browser navigation as the default customer-discovery path.
44	5. Run `python tools/collect_prospects.py --discovery <DISCOVERY.yaml> --product <PRODUCT.yaml> --output-dir <folder>` when file output is needed.
45	6. If the user only names a product or SKU, pass `--product-query` or `--sku` instead of rewriting `PRODUCT.yaml`.
46	7. If the agent has already identified source pages during the task, pass them at runtime with `--source-url` instead of editing `DISCOVERY.yaml`.
47	8. If `discovery_mode` is `native_scrapling_spider`, `scrapling_spider.enabled` is true, or runtime `--source-url` values are provided, `tools/collect_prospects.py` routes to `tools/scrapling_spider_runner.py`, uses Scrapling's native `Spider`, and writes `crawl_report.json`.
48	9. If the agent has MCP access configured, call `tools/scrapling_mcp_server.py` through the `collect_prospects` MCP tool and pass runtime `source_urls` for the same native Spider workflow.
49	10. When neither a configured API, configured `scrapling_spider.source_urls`, nor runtime source URLs are available, create search tasks for Google search results, trade show websites, industry directories, prospect company websites, LinkedIn public summaries, and B2B platform public pages.
50	11. If the tool outputs only `prospect_search_tasks.csv`, stop and report `source_status: "search_tasks_only"`; ask for a collection API, a user-provided list, or permission to process a specific approved source.
51	12. Do not manually browse arbitrary B2B platforms after search tasks are generated.
52	13. Do not return a numbered customer list unless it comes from `prospects.raw.csv`, a user-provided prospect file, or fetched company pages with source URLs.
53	14. If sources cannot be accessed or provide no usable company records, return `source_status: "source_unavailable"` and do not fill the gap with industry knowledge.
54	15. With a configured API, produce `prospects.raw.csv` and `prospects.raw.json`.
55	16. Build keyword groups from product names, HS codes, applications, buyer types, target regions, and channel terms.
56	17. Add exclude terms for jobs, consumer reviews, unrelated retail-only pages, marketplaces without seller websites, and irrelevant industries.
57	18. Keep each candidate tied to a source URL, company-level signal, and reviewable evidence summary.
58	19. When discovery finds a company name or company link, visit the official website or source-linked company page next and run official website contact search before finalizing the row.
59	20. During official website contact search, check homepage, contact, about, team, catalog, and product pages for visible email and phone values.
60	21. For each candidate, record company name, website, country, business type, source URL, evidence summary, risk notes, contact email, contact phone, email result, and phone result.
61	22. If email or phone is not found during discovery, write `contact_email: "没有"`, `contact_phone: "没有"`,

People who liked this also viewed

6 picks

Type