Smb Sales Boost

Rate Limits

• Exports: 1 per 5 minutes
• Email schedule trigger: 1 per 5 minutes
• AI category suggestions: 5 per minute
• AI keyword generation: 5 per minute
• AI auto-refine enable: 5 per minute
• AI auto-refine disable: 60 per minute
• AI auto-refine status: 60 per minute
• AI keyword status: 60 per minute

Rate limit headers are returned on every response: X-RateLimit-Limit, X-RateLimit-Remaining, X-RateLimit-Reset. If rate limited, check the Retry-After header for seconds to wait.

Two Database Types

SMB Sales Boost has two separate databases with different contact information available:

1. home_improvement — Home improvement/contractor businesses with phone numbers, star ratings, review counts, review snippets, profile URLs, and categories
2. other — General newly registered businesses with phone numbers and email addresses, registered URLs, crawled URLs, short/long descriptions, and redirect status

The Home Improvement database provides phone numbers as the primary contact method. The Other database provides both phone numbers and email addresses, making it ideal for cold email and multi-channel outreach campaigns.

Some filter parameters only work with one database type. The user's account has a default database setting. Always check which database the user wants to query.

Core Endpoints

1. Search Leads — GET /leads

The primary endpoint. Translates natural language queries into filtered lead searches.

Key Parameters:

Understanding "Last Updated" — this is critical for finding the freshest leads:

• Home Improvement leads: Last Updated means a new phone number was detected
• Other leads: Last Updated means the primary phone number and/or primary email address was updated
• Both databases also include newly added records in this date
• Many businesses launch a website before adding contact info, so the Last Updated date captures when that information first becomes available — making it the primary way to identify the most actionable leads

Wildcard Keyword Tips:

• Use * to match any characters: "*dental*" matches "dental clinic", "pediatric dentistry", etc.
• Combine wildcards for compound terms: "*auto*repair*" matches "auto body repair", "automotive repair shop", etc.
• Use multiple keyword variations for broader coverage: ["*dental*", "*dentist*", "*orthodont*"]
• Keywords without wildcards still perform substring matching by default

Home Improvement Only:

Other Database Only:

Important: At least one positive filter is required (positiveKeywords or any column-specific include terms).

Response includes: leads array, totalCount, page, limit, databaseType

Lead fields: id, companyName, state, city, zip, phone, email, categories, lastUpdated (phone/email masked for free users). The lastUpdated field indicates when contact information was last detected or updated — this is the best indicator of lead freshness and actionability.

2. Export Leads — POST /leads/export

Export filtered leads as CSV, JSON, or XLSX files.

Request body:

{
  "database": "home_improvement" | "other",
  "filters": { /* same filter params as GET /leads */ },
  "selectedIds": [1, 2, 3],  // alternative to filters
  "formatId": 123  // optional export format template ID
}

Response: files array (with base64-encoded data), leadCount, exportId, databaseType

Rate limited: 1 export per 5 minutes, max 10,000 leads per export.

3. Filter Presets — /filter-presets

• GET /filter-presets — List all saved presets
• POST /filter-presets — Create a preset (requires name and filters object)
• DELETE /filter-presets/{id} — Delete a preset

4. Keyword Lists — /keyword-lists

• GET /keyword-lists — List all keyword lists
• POST /keyword-lists — Create (requires name, optional keywords and negativeKeywords arrays)
• PUT /keyword-lists/{id} — Update
• DELETE /keyword-lists/{id} — Delete

5. Email Schedules — /email-schedules

• GET /email-schedules — List schedules
• POST /email-schedules — Create (requires name, filterPresetId, intervalValue, intervalUnit)
• PATCH /email-schedules/{id} — Update (supports isActive toggle)
• DELETE /email-schedules/{id} — Delete
• POST /email-schedules/{id}/trigger — Manually trigger an active schedule to send immediately (rate limited: 1 per 5 minutes)

6. Export Formats — /export-formats

• GET /export-formats — List custom export formats
• POST /export-formats — Create (requires name, supports fileType, fieldMappings, split settings)
• GET /export-formats/{id} — Get specific format
• PATCH /export-formats/{id} — Update
• DELETE /export-formats/{id} — Delete
• POST /export-formats/{id}/set-default — Set as default

7. Export History — /export-history

• GET /export-history — List past exports (optional limit param, default 50)
• GET /export-history/{id}/download — Re-download (expires after 7 days)

8. AI Features

POST /ai/suggest-categories — Get AI category suggestions based on company profile.

Required: companyName, companyDescription, productService Optional: companyWebsite, smbType, excludeCategories

POST /ai/generate-keywords — Trigger async keyword generation based on profile and target categories.

GET /ai/keyword-status — Check the status of keyword generation jobs. Use this to poll for completion after triggering keyword generation.

AI Auto-Refine — Automatically refine keyword lists using AI:

• POST /ai/auto-refine/enable — Enable auto-refine for a keyword list (requires listId)
• POST /ai/auto-refine/disable — Disable auto-refine for a keyword list (requires listId)
• GET /ai/auto-refine/status — Check auto-refine status (optional listId query param to filter by specific list)

9. Export Blacklist — /export-blacklist

• GET /export-blacklist — List blacklisted entries
• POST /export-blacklist — Add entry (single or batch via entries array)
• DELETE /export-blacklist/{id} — Remove entry

10. Account

• GET /me — Get user profile (subscription, settings, onboarding status)
• PATCH /me — Update profile (firstName, lastName, companyName, companyWebsite)
• GET /settings/database — Check current database type and switch availability
• POST /settings/switch-database — Switch between databases (has cooldown)

Natural Language Translation Guide

When users make natural language requests, translate them into API calls. Use multiple wildcard keyword variations to cast a wider net — keywords are matched via OR logic so more variations means better coverage:

Building API Requests

Use the included smb_api.py script for all API calls. It handles authentication, URL encoding, response parsing, and safe file export in a single reusable file. Do not use shell commands like curl — constructing shell commands from user-provided input risks shell injection vulnerabilities.

Usage

python smb_api.py <API_KEY> <METHOD> <ENDPOINT> [--params '{"key":"value"}'] [--body '{"key":"value"}'] [--output-dir /path/to/dir]

Examples

# Search for med spas in Florida using wildcard keywords (OR logic)
python smb_api.py smbk_xxx GET /leads --params '{"positiveKeywords":"[\"*med*spa*\",\"*medical*spa*\",\"*aesthet*\",\"*botox*\",\"*medspa*\"]","stateInclude":"FL","limit":"25"}'

# Find auto shops in multiple states, exclude franchises
python smb_api.py smbk_xxx GET /leads --params '{"positiveKeywords":"[\"*auto*repair*\",\"*body*shop*\",\"*mechanic*\",\"*tire*\",\"*oil*change*\"]","negativeKeywords":"[\"*franchise*\",\"*jiffy*\"]","stateInclude":"GA,FL,NC,SC,TN","limit":"50"}'

# Search for recently updated dental leads in Texas
python smb_api.py smbk_xxx GET /leads --params '{"positiveKeywords":"[\"*dental*\",\"*dentist*\",\"*orthodont*\",\"*oral*surg*\"]","stateInclude":"TX","lastUpdatedFrom":"2026-02-14"}'

# Get account info
python smb_api.py smbk_xxx GET /me

# Export pet industry leads (files saved automatically to --output-dir)
python smb_api.py smbk_xxx POST /leads/export --body '{"database":"other","filters":{"positiveKeywords":["*pet*groom*","*veterinar*","*dog*train*","*pet*board*"],"stateInclude":"CA,OR,WA"}}'

# AI category suggestions for a fitness equipment distributor
python smb_api.py smbk_xxx POST /ai/suggest-categories --body '{"companyName":"FitPro Supply","companyDescription":"Commercial fitness equipment distributor","productService":"Gym equipment, treadmills, weight systems"}'

# Create a filter preset for bakery/catering leads
python smb_api.py smbk_xxx POST /filter-presets --body '{"name":"NY Bakeries","filters":{"positiveKeywords":["*bakery*","*bake*shop*","*cater*","*pastry*"],"stateInclude":"NY"}}'

# Enable AI auto-refine on a keyword list
python smb_api.py smbk_xxx POST /ai/auto-refine/enable --body '{"listId":42}'

# Check auto-refine status for a specific list
python smb_api.py smbk_xxx GET /ai/auto-refine/status --params '{"listId":"42"}'

# Check keyword generation job status
python smb_api.py smbk_xxx GET /ai/keyword-status

# Manually trigger an email schedule
python smb_api.py smbk_xxx POST /email-schedules/15/trigger

# Delete a filter preset
python smb_api.py smbk_xxx DELETE /filter-presets/42

The script outputs JSON to stdout and rate limit headers to stderr. For export requests, files are automatically saved with sanitized filenames.

Remember:

• Use multiple wildcard keyword variations to cast a wider net (e.g., ["*dental*", "*dentist*", "*orthodont*"] not just ["dental"]) — keywords are matched via OR logic
• Use * for flexible pattern matching: "*auto*repair*" matches "auto body repair", "automotive repair shop", etc.
• JSON array parameters should be serialized as strings inside the --params JSON
• At least one positive filter is required for lead searches
• Check which database the user needs before applying database-specific filters
• Home Improvement database provides phone numbers; Other database provides phone numbers and email addresses
• Phone and email are masked for free-tier users
• Present results in a clean, readable table format

Security

This skill addresses two specific agent execution risks: shell injection from constructing CLI commands with user input, and arbitrary file writes from unsanitized API-provided filenames.

Shell injection prevention: The smb_api.py script uses Python's requests library for all HTTP calls. User-provided search terms, locations, and other inputs are passed as structured function arguments — never interpolated into shell command strings. This eliminates the shell injection vector that exists when agents construct curl commands from user input.

Path traversal prevention in exports: The /leads/export endpoint returns base64-encoded files with an API-provided fileName field. A malicious or corrupted filename (e.g., ../../etc/passwd) could write files to arbitrary locations. The script enforces three safeguards:

1. Basename extraction: os.path.basename() strips all directory components — ../../etc/passwd becomes passwd
2. Extension validation: Only .csv, .json, and .xlsx extensions are allowed; anything else defaults to .csv
3. Scoped output directory: Files are written only to the designated output directory (/mnt/user-data/outputs/ by default), never to user-specified or API-specified paths

API key handling: The key is passed as a CLI argument and sent only in the Authorization header. It is never logged, written to files, or included in error output.

Error Handling

All errors return: { "error": "error_code", "message": "Human-readable message" }