# SDKGPT

> SDKGPT is a mobile app SDK intelligence platform. It scrapes APKs from app stores, parses SDK signals (.so/Service/Activity/DEX), matches them against the LibChecker knowledge base, and outputs SDK intelligence. Covers 1686 SDKs, 57000+ DEX package name signals, 127K Android apps, and 828 companies.

SDKGPT helps SDK vendors find customers, investors do market analysis, and security researchers batch-screen apps. Browsing data is free forever; combo exploration for lead finding, full market share, and export are paid B2B features.

The data pipeline has 5 stages: scrape app stores → download APK → parse .so/DEX/Service/Activity → match LibChecker SDK rules → output SDK intelligence.

## Product

- [Home](https://sdkgpt.com/): SDKGPT mobile app SDK intelligence platform homepage, 5-stage pipeline monitor
- [Apps Directory](https://sdkgpt.com/apps): Browse 127K Android apps by category, view each app's SDK tech stack
- [SDKs Directory](https://sdkgpt.com/sdks): Browse 1686 SDKs, view SDK details, parent company, market share
- [Companies Directory](https://sdkgpt.com/companies): Browse 828 companies, view their SDK products
- [Combo Explorer](https://sdkgpt.com/explore): Advanced filtering — uses X not Y, multi-SDK combos, size tiers, batch package name import
- [Market Share](https://sdkgpt.com/market-share): SDK market share rankings, penetration trends, category rankings
- [SDK Categories](https://sdkgpt.com/categories): Browse SDKs by category (ads, push, analytics, maps, etc.)
- [Recent SDKs](https://sdkgpt.com/recent-sdks): Latest discovered SDK activity
- [Pricing](https://sdkgpt.com/pricing): Free to start, Pro monthly subscription for full results and export

## User Access

- [Login](https://sdkgpt.com/login): SDKGPT user login
- [Sign Up](https://sdkgpt.com/signup): Sign up for SDKGPT for free
- [Feedback](https://sdkgpt.com/feedback): Product feedback and feature requests

## Core Data

- 1686 SDK rules (covering .so/DEX/Service/Activity signal types)
- 57000+ DEX package name signals
- 127K Android apps (Huawei + MyApp stores)
- 828 companies (normalized via package prefix → domain → host_domain → canonical_name)

## Data Pipeline (5 stages)

1. **Scrape app stores**: Huawei + MyApp stores, resumable crawling + keyword auto-expansion
2. **Download APK**: aria2c multi-threaded download, resumable
3. **Parse APK**: androguard parses .so/DEX/Service/Activity signals, subprocess isolation prevents OOM (each APK gets a fresh subprocess, RLIMIT_AS 8GB hard limit)
4. **Match SDK rules**: LibChecker original logic (`==` || `startsWith(prefix+".")`), covers .so/DEX/Service/Activity signals
5. **Output SDK intelligence**: Written to database, frontend display + API queries

## SDK Signal Types

SDKGPT parses 4 signal types from APKs to identify SDKs:

- **.so signal (signal_type=0)**: Native library filenames (e.g. `libBaiduMap.so`). Shallow extraction gets filename, deep extraction uses pyelftools to extract JNI function names and internal strings from .so binary
- **DEX package signal (signal_type=1)**: Java/Kotlin package name prefixes (e.g. `com.umeng.analytics`). 57000+ package signals covered
- **Service/Activity declaration (signal_type=2)**: Service and Activity component names declared in AndroidManifest.xml
- **URL signal (signal_type=10)**: http/https URL hosts extracted from DEX strings (e.g. `ulogs.umeng.com`). Used to discover pure REST API integrations, cloud services/SaaS with no SDK package name trace

## SDK Matching Rules

Matching follows LibChecker's original logic:

- Package name: `==` (exact match) or `startsWith(prefix+".")` (prefix + dot, not simple startsWith)
- .so: filename exact match or prefix match
- Service/Activity: component name exact match or prefix match
- URL: host suffix match against `sdk_company.domain`

Match results have two levels:
- **confirmed signal**: Passed rule matching, `is_confirmed=true`
- **potential signal**: Not yet rule-matched, `is_confirmed=false`, upgraded by LLM final judgment

## Target Users & Use Cases

### SDK Vendors (BD/Sales)

- **Find leads**: Combo explorer "uses X not Y" — find apps using competitor SDK but not yours
- **Filter leads**: Size tiers (large/medium/small/long tail) for prioritization
- **Monitor competitors**: Track competitor SDK market share changes and new app adoption
- **Batch screening**: Batch import 100 package names to query SDK tech stacks

### Investors

- **Market analysis**: SDK market share rankings, penetration trends, category rankings
- **Company research**: View company SDK products and market share
- **Trend discovery**: Recent SDK activity, discover emerging technologies

### Security Researchers

- **Batch screening**: Combo explorer to filter apps by SDK, identify security concerns
- **App tech stacks**: View complete SDK tech stack details for individual apps

## Combo Explorer Filters

Combo explorer is SDKGPT's core paid feature, supporting multi-dimensional combined filtering:

### SDK Filters (multi-group + mixed operators)

Supports up to 20 SDK filter groups, each with multiple SDKs combined by operator:

- **any (union)**: App uses any SDK in the group (`sdks IN [A,B,C]`)
- **all (intersection)**: App uses all SDKs in the group (`sdks CONTAINS [A,B,C]`)
- **not_any (exclusion)**: App uses none of the SDKs in the group (`NOT (sdks IN [A,B,C])`)

Multiple groups are AND-combined. Typical usage:
- "Uses Bugly but not OkHttp" = g0(any:Bugly) + g1(not_any:OkHttp)
- "Uses both Bugly and OkHttp" = g0(all:Bugly,OkHttp)
- "Uses Bugly or OkHttp, but not Retrofit" = g0(any:Bugly,OkHttp) + g1(not_any:Retrofit)

URL parameter format:
- Legacy compat: `sdks=Bugly,OkHttp&sdk_op=any` (equivalent to g0)
- Legacy compat: `sdks_not_any=Retrofit` (equivalent to g1 not_any)
- New: `sdks_g0=Bugly,OkHttp&sdk_op_g0=any&sdks_g1=Retrofit&sdk_op_g1=not_any`

### Size Tiers

By download_count:
- **Large**: ≥ 10M downloads
- **Medium**: 100K - 10M downloads
- **Small**: 10K - 100K downloads
- **Long tail**: < 10K downloads

URL param: `size_bucket=large|medium|small|long_tail`

### Region Filter

Filter by app developer's province/city. Provinces support both full and short Chinese names.

URL params: `province=Guangdong,Zhejiang&city=Hangzhou`

### Full-text Search

Search by app name or package name.

URL param: `q=WeChat` or `q=com.tencent.mm`

### Batch Package Import

Search box auto-detects comma/newline-separated package names (max 100), switches to batch query mode.

URL param: `bundle_ids=com.tencent.mm,com.tencent.mobileqq`

## API Endpoints

All APIs return JSON, require Supabase JWT Bearer token for Pro features.

### Combo Explorer API

- [GET /api/search/apps](https://sdkgpt.com/api/search/apps): Combo explorer query. Params: `sdks_g0`, `sdk_op_g0`, `province`, `city`, `size_bucket`, `q`, `bundle_ids`, `page`, `limit`. Free users get first 10 results, Pro users full pagination
- [GET /api/search/apps/export](https://sdkgpt.com/api/search/apps/export): Export CSV/JSON. Same params + `format=csv|json`. Pro only

### App Detail API

- [GET /api/apps/[appId]/detail](https://sdkgpt.com/api/apps/): App details. `appId` = bundle_id (e.g. `com.tencent.mm`). Returns app info + SDK tech stack list

### SDK APIs

- [GET /api/sdks/types](https://sdkgpt.com/api/sdks/types): SDK category list (ads/push/analytics/maps etc.)
- [GET /api/sdk-filters](https://sdkgpt.com/api/sdk-filters): SDK filter data (for combo explorer SDK picker)
- [GET /api/sdk-taxonomy](https://sdkgpt.com/api/sdk-taxonomy): SDK taxonomy

### Market Share API

- [GET /api/market-share](https://sdkgpt.com/api/market-share): SDK market share ranking. Params: `subcategory_id`, `limit`. Free users get TOP 10 preview, Pro users full ranking (max 1000)

### Recent SDKs API

- [GET /api/recent-sdks](https://sdkgpt.com/api/recent-sdks): Recently confirmed SDKs. Params: `limit` (default 50, max 100)

### Company APIs

- [GET /api/companies/list](https://sdkgpt.com/api/companies/list): Company list
- [GET /api/companies/[id]/clients](https://sdkgpt.com/api/companies/): Company clients (apps using this company's SDKs)

### Payment & Subscription APIs

- [POST /api/payments/wechat/order](https://sdkgpt.com/api/payments/wechat/order): Create WeChat Pay order
- [GET /api/payments/wechat/query](https://sdkgpt.com/api/payments/wechat/query): Query payment status
- [GET /api/payments/history](https://sdkgpt.com/api/payments/history): Payment history
- [POST /api/subscriptions/cancel](https://sdkgpt.com/api/subscriptions/cancel): Cancel subscription (remains active until period end)
- [POST /api/payments/backfill](https://sdkgpt.com/api/payments/backfill): Backfill payment records

### User APIs

- [GET /api/saved-searches](https://sdkgpt.com/api/saved-searches): Saved search queries (Pro only)
- [POST /api/feedback](https://sdkgpt.com/api/feedback): Submit feedback
- [POST /api/feedback/[id]/vote](https://sdkgpt.com/api/feedback/): Vote on feedback
- [DELETE /api/account/delete](https://sdkgpt.com/api/account/delete): Delete account

## Data Schema (Key Fields)

### App

| Field | Description |
|-------|-------------|
| `app_id` | bundle_id (Android package name, e.g. `com.tencent.mm`) |
| `name` | App name |
| `icon_url` | App icon URL |
| `category` | App category (Games/Social/Tools/Shopping etc.) |
| `developer` | Developer name |
| `download_count` | Download count |
| `rating` | Rating |
| `province` | Developer's province |
| `city` | Developer's city |
| `icp_entity` | ICP registration entity |
| `operator` | Operating entity |
| `restrict_level` | App restriction level |
| `sdks` | Array of identified SDK names |
| `sdks_count` | SDK count |

### SDK

| Field | Description |
|-------|-------------|
| `sdk_name` | SDK name (e.g. `Bugly`) |
| `sdk_cn_name` | SDK Chinese name |
| `slug` | URL slug |
| `category_name` | Category name |
| `subcategory_name` | Subcategory name |
| `company_name` | Parent company |
| `app_count` | Number of apps using this SDK |
| `penetration_rate` | Penetration rate |
| `match_type` | Match type (.so/DEX/Service/Activity/URL) |

### Company

| Field | Description |
|-------|-------------|
| `name` | Company name |
| `domain` | Official domain |
| `canonical_name` | Normalized name |
| `sdk_count` | Number of SDKs |

Company normalization: package prefix → domain → host_domain → canonical_name → create new. Package prefix is most accurate, only looks at DEX confirmed signals.

## Pricing

### Free

- Browse SDK / company / app directories
- View individual app SDK tech stacks
- View SDK / company detail pages
- Combo explorer: total count + 10 result preview
- SDK market share TOP 10 preview
- SDK category ranking preview
- Global search (⌘K)

### Pro

#### Combo Explorer Platform

- Full lead lists (no 10 result limit)
- Advanced filtering: uses X not Y / multi-SDK combos
- Size tiers (large / medium / small / long tail)
- Batch package import (100)
- Save search queries
- CSV / JSON export

#### Market Intelligence

- Full SDK market share ranking
- SDK penetration trends (quarterly)
- Company SDK client reports (add/remove changes)

#### Team

- 3 user seats
- Email support

## FAQ

- **What's free?** Browsing data is completely free. Paid features are B2B: full lead lists, advanced filtering, export, full market share rankings.
- **Combo explorer free vs Pro?** Free gets total count + 10 preview results. Pro gets full list + pagination + export + saved searches + batch import.
- **Can I cancel anytime?** Monthly subscription cancels anytime, remains active until billing period ends, then downgrades to free.
- **Annual discount?** Annual billing gets 20% off. Contact sales@sdkgpt.com.
- **Data coverage?** Currently covers 127K Android apps, 1686 SDKs, 828 companies. Expanding to global Google Play and Apple App Store.
- **Is SDK matching accurate?** Uses LibChecker original logic (`==` || `startsWith(prefix+".")`), covers .so/DEX/Service/Activity signals. URL signals use host suffix matching + LLM final judgment.
- **How often is data updated?** Crawler runs continuously, app data updated daily, SDK rule library continuously expanded.