Roadmap
Our goal: match every feature in the Agno framework so you can visually build anything Agno supports. Here's what's shipped, what's in progress, and what's coming next.
Releases
What's next
We're expanding the tool ecosystem from 16 to 120+ tools with a categorised browser, and adding Knowledge Base nodes with RAG support — connecting PDFs, websites, and databases directly to your agents on the canvas.
Live features
Tier 1 Model Providers
Most-requested providers: Claude, Gemini, DeepSeek, Groq, Mistral, and xAI — each with API key management and correct Python export.
- Provider selector dropdown in ConfigPanel
- Backend: dynamic model instantiation by provider + model ID
- API key management per provider in UI
- Python code export with correct provider imports
- Type support for provider + model pairs
Tier 2 Model Providers
Cohere, Together AI, Fireworks AI, Perplexity, OpenRouter, Azure OpenAI, and AWS Bedrock. Multi-key UI for providers that need more than one credential.
- Cohere (Command R+, Command R, Command Light)
- Together AI (Llama 3, Mixtral)
- Fireworks AI (Llama 3.1, Mixtral)
- Perplexity (Sonar Large / Small online)
- OpenRouter (unified API for 100+ models)
- Azure OpenAI (multi-key: API key + endpoint)
- AWS Bedrock (multi-key: access key + secret + region)
UI/UX Overhaul
Dark mode with 90+ design token replacements, mobile-responsive layout with slide-over panels, onboarding tour, canvas empty state, and AI Elements chat rewrite.
- Dark mode overhaul — replaced 90+ hard-coded colors with design tokens
- Mobile-responsive layout with ConfigPanel slide-over drawer
- Interactive onboarding tour using Driver.js
- Canvas empty state with call-to-action
- AI Elements chat UI rewrite (PromptInput, Attachments, Markdown)
- Hide Save button and ConfigPanel when canvas is empty
Accessibility & Hardening
Keyboard navigation, form label associations, API route validation, import error messages, and API key stripping from exported workflows.
- Keyboard navigation for sidebar and form labels (a11y)
- API route validation with content-type and body size checks
- User-facing import error messages
- Strip API keys from exported workflow JSON
- ChatPanel memoization and callback stability
Multimodal Attachments
Attach images, PDFs, DOC, and PPT files in chat. Agents can see and describe images, with grid-style preview cards and file upload guardrails.
- Image, PDF, DOC, PPT file attachment in chat
- Grid-style 96x96px attachment preview cards
- File upload guardrails (max 5 files, 5MB each)
- Base64 data URL decoding for Agno compatibility
- Frontend-to-backend attachment pipeline (end-to-end)
Reasoning & Tool Call UI
Upgraded reasoning display with streaming chain-of-thought, auto-open/close, duration tracking, and tool call visualization with status badges.
- Streaming-aware reasoning with auto-open/close
- Duration tracking ("Thought for 3 seconds")
- Shimmer animation while thinking
- Tool call status badges (Running, Completed, Error)
- Rich Streamdown markdown rendering (code, math, mermaid)
Tool Metadata Enhancement
Search tags, documentation URLs, built-in status flags, and improved descriptions for all 16 tools. Foundation for search/filter UI.
- Added tags, documentationUrl, isBuiltIn to ToolConfig
- All 16 tools enriched with search tags
- Documentation URLs linking to docs.agno.com
- Built-in vs pip-installable tool flags
- Improved tool descriptions