Changelog

Add HTTPS Request Methods.

Added GET, POST, PATCH, DELETE, OPTIONS request methods into Flask endpoints, updated Test Request tab.

Previously, all requests were defined through the GET request method, even for applications where other request methods are more appropriate.

Preset test requests based on ASM trees

Servera provides you with preset request routes and parameters for much easier testing using the generated ASM trees.

Complete change of LLM logic to a chat-based conversational develpoment

Collaborates with users to create detailed functions and PRD for each route before creating. Saves chat states for efficiency.

Upgrade to Gemini 2.0 Flash for better generation accuracy and speed.

Export function is still using the legacy Gemini 1.5 flash.

Added and Tested More External Integrations

Added and did basic functionality testing for integrations for stock data (Yahoo Finance), geolocation and points of interest (geopy, OpenStreetMap), information and data (Wikipedia), and external APIs.

Servera will soon migrate to the AI autofill as the default way to generate. For now, it's best to use AI Autofill as it knows what features Servera has access to.

Fixed an issue with user-created AI agents hallucinating

Cleared chain of thought for each separate user who accessed the API to ensure reasoning is reset per request

This should fix an issue with AI caching previous responses, leading to an eventual persistent error with all AI agents.

Route visualization

Added a routes tab, where the user can see the input/output of each function, and how each route is being serviced.

All users can access this visualizer through the Routes tab.

Endpoint modifications

Fixed an issue where modifying your endpoint would deregister user tunnels.

Previously, this would unlink the tunnel to the process. We have fixed this issue, and endpoint modification and deletion should fully work as intended now.

Migrate proxy

Moved temporary cloudflare proxies to Servera's own infrastructure for permanent responsive APIs.

Agentic AI Integration

Users can build wrapper apps using Gemini AI models, as well as build their own AI agent through a system prompt.

This feature is still in development, and there could be many bugs. If you run into an issue, please report it through our feedback tab in our header. Thanks for using Servera!

Prompt exports

All users have the ability to have AI generate a message to provide context on all routes.

Users can provide this message to any AI model of their choice, which can recognize and understand what each route does.

WIP: Python libraries

Currently work in progress as of March 12: Implement python library presets for large data set applications, machine learning, LLMs, and data visualization, etc.

Status: Implementing

Database creation and persistence implementation

Users can have the option to create a persistent SQLite database.

The database persists and keeps its data even if Servera has an outage, or if the user redeploys their endpoint by modifying the codebase.

Test request polling time tracking

Sent test requests now show the amount of time (ms) it took for the response to be returned.

Also, error codes for specific errors are also displayed.

UI/UX update and design

Updated the logo, UI layout and design, and landing page

Clerk Authentication

Added user accounts using Clerks OAuth support.

Added multi-user endpoints support

Implemented separated Gunicorn HTTP server workers for each user.

Contextual AI and chat console

Users can now chat with the AI directly and modify their endpoints live.

The AI can read the entire backend program and provide changes, comment on code, and reply to the user effectively, with memory of prompts.

Migrate to Vercel and Cloudflare

Moved servera.dev to be hosted by Vercel and Cloudflare

Initial release

Created and worked on the development of dynamic API routes, and generating backend logic with AI.

Developed the idea of Servera, the use case, roadmap, etc.