prospector（prospector 搜索）

Prospector

Find leads matching your ICP via Exa company search + Apollo contact enrichment.

Prerequisites

Run /prospector:setup first to configure your API keys:

• Exa (required): https://exa.ai - company discovery
• Apollo (required): https://apollo.io - contact enrichment
• Attio (optional): https://attio.com - CRM sync

You can also set keys via environment variables:

• PROSPECTOR_EXA_API_KEY
• PROSPECTOR_APOLLO_API_KEY
• PROSPECTOR_ATTIO_API_KEY (optional)

Usage

Setup (one-time)

/prospector:setup

Collects and validates API keys, stores securely in ~/.config/prospector/config.json.

Find Leads

/prospector

Asks ICP questions, searches Exa, enriches via Apollo, outputs CSV to Desktop.

Main Command: /prospector

When the user invokes /prospector, follow this workflow:

Step 1: Check Config or Env Vars

First, verify env vars or config exist:

python3 -c "
import json
import os
from pathlib import Path
config_path = Path.home() / '.config' / 'prospector' / 'config.json'
env_exa = bool(os.getenv('PROSPECTOR_EXA_API_KEY'))
env_apollo = bool(os.getenv('PROSPECTOR_APOLLO_API_KEY'))
env_attio = bool(os.getenv('PROSPECTOR_ATTIO_API_KEY'))
if not config_path.exists():
    print('NOT_FOUND')
else:
    with open(config_path) as f:
        config = json.load(f)
    print('FOUND')
    print(f'exa: {bool(config.get(\"exa_api_key\"))}')
    print(f'apollo: {bool(config.get(\"apollo_api_key\"))}')
    print(f'attio: {bool(config.get(\"attio_api_key\"))}')
print(f'env_exa: {env_exa}')
print(f'env_apollo: {env_apollo}')
print(f'env_attio: {env_attio}')
"

If NOT_FOUND and env vars are not set, tell user to run /prospector:setup first.

Step 2: Ask ICP Questions

Use AskUserQuestion to collect ICP criteria in order:

Question 1: Industry

header: "Industry"
question: "What industry are you targeting?"
options:
  - label: "SaaS"
    description: "Software as a Service companies"
  - label: "Fintech"
    description: "Financial technology companies"
  - label: "Healthcare"
    description: "Healthcare and health tech"
  - label: "E-commerce"
    description: "Online retail and marketplaces"
  - label: "AI/ML"
    description: "Artificial intelligence and machine learning"
  - label: "Any"
    description: "No industry filter"
multiSelect: false

Question 2: Company Size

header: "Size"
question: "What company size are you targeting?"
options:
  - label: "1-10"
    description: "Early stage startups"
  - label: "11-50"
    description: "Seed to Series A"
  - label: "51-200"
    description: "Series A to B"
  - label: "201-500"
    description: "Growth stage"
  - label: "500+"
    description: "Enterprise"
  - label: "Any"
    description: "No size filter"
multiSelect: false

Question 3: Funding Stage

header: "Funding"
question: "What funding stage are you targeting?"
options:
  - label: "Pre-seed"
    description: "Pre-product market fit"
  - label: "Seed"
    description: "Building initial product"
  - label: "Series A"
    description: "Scaling product"
  - label: "Series B+"
    description: "Growth and expansion"
  - label: "Any"
    description: "No funding filter"
multiSelect: false

Question 4: Geography

header: "Geography"
question: "What geography are you targeting?"
options:
  - label: "United States"
    description: "US-based companies"
  - label: "Europe"
    description: "European companies"
  - label: "Global"
    description: "Worldwide"
  - label: "Any"
    description: "No geography filter"
multiSelect: false

Question 5: Keywords (optional)

header: "Keywords"
question: "Any specific keywords that should appear in company descriptions? (optional)"
options:
  - label: "Skip"
    description: "No keyword filter"
  - label: "Enter keywords"
    description: "I'll type specific keywords"
multiSelect: false

If "Enter keywords", ask for the text input.

Question 6: Contact Count

header: "Count"
question: "How many contacts do you want to find?"
options:
  - label: "25"
    description: "Quick search, lower API usage"
  - label: "50"
    description: "Balanced (recommended)"
  - label: "100"
    description: "Larger batch, more API usage"
multiSelect: false

Step 3: Run Search

Execute the Python script with the collected ICP:

cd [skill_directory]/scripts
python3 -c "
from prospector import run_search, export_csv, Config

icp = {
    'industry': '[INDUSTRY]',
    'company_size': '[SIZE]',
    'funding_stage': '[FUNDING]',
    'geography': '[GEOGRAPHY]',
    'keywords': '[KEYWORDS or empty string]'
}

leads = run_search(icp, num_contacts=[COUNT])
if leads:
    path = export_csv(leads)
    print(f'SUCCESS: {len(leads)} leads saved to {path}')
else:
    print('NO_RESULTS: No leads found matching criteria')
"

Replace placeholders with actual values from questions.

Step 4: Attio Sync (if configured)

If Attio is configured and leads were found, ask:

header: "Attio"
question: "Sync leads to Attio CRM?"
options:
  - label: "Yes"
    description: "Sync companies and contacts to Attio"
  - label: "No"
    description: "Just keep the CSV"
multiSelect: false

If yes:

cd [skill_directory]/scripts
python3 -c "
from prospector import sync_to_attio, Config, Lead
import json

# Load leads from the CSV we just created (or pass them directly)
# For simplicity, re-run with sync
leads = [...]  # Pass leads from previous step

companies, people = sync_to_attio(leads)
print(f'SYNCED: {companies} companies, {people} contacts')
"

Step 5: Report Results

Tell the user:

• How many leads were found
• Where the CSV was saved
• If Attio sync was done, how many records were synced

Error Handling

• Config not found: Tell user to run /prospector:setup
• Invalid API key: Tell user which key failed, suggest re-running setup
• No results: Suggest broadening ICP criteria
• Partial failures: Report what succeeded, warn about failures