API Testing
Run benchmark requests against OpenAI-compatible endpoints and capture TTFT, total latency, output speed, failure causes, and usage-source attribution.
Local-first AI API testing desktop
AI API testing and configuration desktop
Windows 1.0.0 is available; macOS and Linux are not yet open for download
LOCAL-FIRST API WORKSTATION
shinning
API testing workspace
API testing, API cards, providers, prompts, and Agent Skills are managed in separate desktop layers.
Run benchmark requests against OpenAI-compatible endpoints and capture TTFT, total latency, output speed, failure causes, and usage-source attribution.
Store API Key, Base URL, model, provider, tags, and notes as locally encrypted configuration cards.
Maintain official, relay, local, and custom providers with consistent endpoint, model, and parameter presets.
Separate benchmark prompts from user prompts so test baselines and day-to-day templates remain distinct.
Organize Skill Packs and documentation Skills so agents can reuse explicit API instructions, input requirements, and operating boundaries.
Time to first token, used to evaluate response start-up latency.
The full duration from request dispatch to completed response.
Generated token speed calculated from output tokens and elapsed time.
Identifies provider-reported usage, local estimates, mixed data, or unknown sources.
Local security
Desktop credentials remain on the user device by default. The official website publishes product information, download entry points, release manifests, and legal documents. It does not provide API key cloud sync or hosted proxy requests.
API keys remain on the user's device by default.
System-level local encryption is used instead of plaintext JSON storage.
API card data is re-encrypted with a user-provided password before migration.
Enter API Key, Base URL, model, provider, tags, and notes, then save them as a locally encrypted card.
Use benchmark prompts to measure TTFT, total latency, output speed, failure cause, and usage-source attribution.
Keep provider presets, model settings, benchmark prompts, and user prompts in separate layers.
Organize Skill Packs and documentation Skills so agents read clear interface instructions and operating boundaries.
No. The official website does not receive desktop API keys. Keys remain on the user's device by default and are stored through system-level local encryption.
No. ShnDock labels provider usage, local estimates, mixed data, and unknown sources. Results are for diagnosis and comparison, not billing.
No. The official website does not provide API key cloud sync. Cross-device migration is handled through password-encrypted .sapi export and import.
No. This repository contains the official website only and does not imply that the ShnDock desktop source code is open source.