Skip to main content

What's New in Version 0.670.0

Released on February 12, 2026

This release introduces powerful new capabilities with Agent Skills for custom agent abilities, Structured JSON Output for schema adherence, enhanced Orchestrator planning with predefined schemas, improved agentic internet search capabilities, expanded hook environment variables, and API enhancements for artifact management! 🚀

🧠 Agent Skills: Custom Capabilities for Any Agent​

Empower Your Agents with Custom Skills​

We're excited to introduce Agent Skills - a revolutionary way to extend any agent's capabilities by creating custom instructions directly within the Settings/Tools section. This feature allows you to define specialised behaviors and abilities that can be assigned to any agent in your workspace.

Key Features:

  • Custom Instruction Creation - Define specific skills and capabilities through intuitive custom instructions
  • Universal Assignment - Apply created skills to any agent across your workspace
  • Settings/Tools Integration - Centralised management of agent capabilities within the familiar tools section
  • Flexible Configuration - Tailor skills to match your specific use cases and workflows

This feature transforms agents from static configurations into dynamically adaptable assistants that can be customized for any task or domain requirement.

📋 Structured JSON Output for Schema Compliance​

Declarative JSON Structure for Operator, Programmer & Assistant Agents​

We now offer Structured JSON Output for Operator, Programmer, and Assistant agents - designed to consistently produce declarative JSON structures that adhere to specific schemas. This is particularly valuable in workflow settings where strict schema adherence is critical.

Capabilities:

  • Schema-Compliant Output - Agents now generate JSON that strictly follows predefined schemas
  • Workflow Integration - Perfect for pipeline and automation scenarios requiring structured data
  • Advanced Settings Access - Enable this functionality under the Advanced settings of your agent
  • Type Safety - Consistent, predictable output format for reliable downstream processing

Whether you're building data pipelines, API integrations, or automated workflows, Structured JSON Output ensures your agents deliver precisely formatted data every time.

🧙 Orchestrator Predefined Plan Schema​

Deterministic Planning Meets Agentic Flexibility​

Orchestrator agents can now leverage predefined plan schemas - enabling deterministic planning structures while maintaining the flexibility to adapt based on various scenarios during agent prompting. This represents the perfect balance between rigid workflows and adaptive agentic behavior.

Enterprise Grade Capabilities:

  • Predefined Plan Schema - Set structured planning templates for consistent execution patterns
  • Adaptive Adjustments - Agents can still dynamically adjust plans based on scenario variations
  • Workflow-Agent Sweet Spot - Combines the reliability of workflows with the intelligence of agents
  • Enhanced Performance - Significantly boosted speed and reliability when performing plans dynamically
  • Better Steerability - Improved control over agentic behaviors during plan execution

This enhancement delivers the best of both worlds: predictable planning structures with intelligent adaptability for complex multi-agent orchestration.

🔍 Enhanced Agentic Internet Search Capabilities​

Smarter Web Queries and Data Extraction​

We've made significant enhancements to agentic capabilities specifically for handling internet searches. These improvements span both querying capabilities and data extraction from webpages with intelligent filtering and retrying mechanisms.

Search Improvements:

  • Advanced Query Capabilities - More intelligent and context-aware search query formulation
  • Smart Data Extraction - Enhanced ability to parse and extract relevant information from webpages
  • Intelligent Filtering - Automatic filtering of irrelevant or low-quality search results
  • Retry Mechanism - Smart retrying for failed queries or incomplete data extractions
  • Robust Error Handling - Better recovery from transient network issues or unavailable sources

Your agents now perform more reliable and comprehensive internet research, delivering higher quality information with fewer interruptions.

🔗 Hook Environment Variable Expansion​

Extended Context for Hooks​

Hooks now expose additional contextual information through environment variables, providing deeper integration capabilities for your custom logic.

New Environment Variables:

  • chat_id - Access the unique identifier of the current conversation
  • workspace_id - Reference the workspace context for multi-tenant scenarios
  • message_id - Track individual messages within the conversation flow

These additions enable hooks to make more informed decisions based on conversation context, workspace membership, and message lineage - perfect for building sophisticated integration logic.

✨ Improved Title Generation UX​

Instant Context Awareness​

We've optimized the title generation process to occur at the beginning of the agent reply rather than at the end. This seemingly small change delivers a noticeably better user experience when interacting with agents.

UX Enhancement:

  • Immediate Context Setting - Users see the conversation context established right from the start
  • Better Flow - More natural conversation progression with early context awareness
  • Reduced Waiting - No delay for title generation at conversation conclusion

This enhancement creates more responsive and intuitive agent interactions from the very first moment.

📮 API Expansion: Artifacts Retrieval​

Fetch All Artifacts by Conversation​

We've expanded our API offering to allow users to easily retrieve all artifacts from any conversation using the chat_id parameter.

New API Capability:

  • Artifact Retrieval by Chat ID - Fetch complete artifact collections from any agent conversation
  • Comprehensive Access - Access files, documents, and outputs generated during conversations
  • Simplified Integration - Easy retrieval for backup, audit, or downstream processing needs

This enhancement makes it simpler than ever to manage and access the valuable outputs generated by your agent conversations.

📋 Summary​

This update includes:

  • Agent Skills - Custom instructions assignable to any agent via Settings/Tools section
  • Structured JSON Output - Schema-compliant JSON generation for Operator, Programmer, and Assistant agents
  • Orchestrator Predefined Plan Schema - Deterministic planning with flexible agentic adaptation
  • Enhanced Internet Search - Smarter queries, better data extraction, and intelligent retry mechanisms
  • Hook Environment Variables - New chat_id, workspace_id, and message_id exposure
  • Title Generation UX - Titles now generated at conversation start for better user experience
  • Artifacts API - Retrieve all conversation artifacts via chat_id

Thank you for using ToothFairy! We're constantly working to improve your experience. If you have any feedback or encounter any issues, please don't hesitate to reach out to our support team at support@toothfairyai.com.

Keep building magic with AI! 🧙‍♀️✨

v0.670.0