Skip to main content
Product Updates

Changelog

New features, improvements, and fixes shipped to Agentlify.

v1.0.0

Initial Release
February 2025

Agentlify is Live

The first public release of Agentlify. Deploy agents like functions, call them like models. Steps, skills, streaming — built in.

Agent Platform

  • Deploy AI agents with a single API call via the dedicated /api/agents endpoint
  • Multi-step agent workflows with composable steps: Research → Plan → Execute → Review
  • Real-time streaming of tool calls, step outputs, and reasoning tokens
  • Agent triggers — invoke agents via HTTP webhooks with Bearer token auth
  • Agent scheduling for recurring tasks and automated pipelines
  • Forkable agent templates — one-click to deploy or customize

Smart Router

  • 35+ models routed across OpenAI, Anthropic, Google, Mistral, and more
  • Per-request model selection optimized for cost, latency, quality, and carbon
  • Configurable optimization weights with preset profiles (balanced, cost-saver, quality-first, low-latency)
  • Automatic fallback chains — if a model fails, the next best is tried instantly
  • OpenAI SDK compatible — change your baseURL, keep your code

Skills & Knowledge

  • Skills library with reusable skill packs agents can reference
  • Custom skill creation with structured knowledge and when-to-use triggers
  • Knowledge bases — upload documents, auto-embed, and attach to agents
  • Skill and knowledge versioning for reproducible agent behavior

Developer Experience

  • agentlify npm package — install and call agents from any JS/TS project
  • OpenAI SDK drop-in compatibility for both router and agent endpoints
  • Python, cURL, and JavaScript examples in every docs page
  • Interactive playground for testing agents and routers in the dashboard
  • API key management with mp_ prefixed keys and per-key usage tracking

Dashboard & Analytics

  • Full dashboard for managing agents, routers, skills, and knowledge bases
  • Request-level analytics with routing decisions, latency, and cost breakdown
  • Agent run history with step-by-step execution traces
  • Credit balance, auto-refill settings, and transaction history

Billing & Plans

  • Pay-per-token pricing at $0.50 per 1M platform tokens
  • Provider costs passed through at cost with zero markup
  • Credit-based system with subscription plans (Agent Starter & Agent Pro)
  • Stripe-powered card management and auto-refill

Infrastructure

  • Firebase-hosted with Cloud Functions for serverless agent execution
  • SSE streaming over HTTP for real-time agent output
  • Rate limiting and request validation on all endpoints
  • Carbon observability — CO₂e tracking per request based on model and region

Documentation

  • Comprehensive docs covering getting started, API reference, and SDK guides
  • Agent creation, triggers, schedules, skills, steps, and tools docs
  • Streaming, error handling, and cost optimization guides
  • Interactive code examples with copy-to-clipboard in JS, Python, and cURL

This is the beginning. More updates coming soon.