Releases: crestalnetwork/intentkit
Releases · crestalnetwork/intentkit
Release v0.11.10
Release v0.11.10
New Features
- Introduced a new abstract logo design for the frontend
Improvements & Bug Fixes
- Improved blockchain transaction reliability by fixing nonce management issues
- Fixed a bug in the asynchronous Web3 client to ensure smoother network interactions
- Added missing configuration files for improved test suite organization
Full Changelog: v0.11.9...v0.11.10
v0.11.9
v0.11.9
New Features
- Added support for setting per-token spending limits on Safe wallets — agents can now have independent spending limits for any ERC20 token, not just USDC
Improvements & Bug Fixes
- Refactored Safe spending limit internals to share a unified implementation
v0.11.8
New Features
- Update LLM reasoning effort config and add Google Vertex AI support
- Add support for Qwen 3.5
- Improve side bar UI
- Automatically generate chat titles
Improvements & Bug Fixes
- Refactor model provider implementation
- Fix chat avatar display issues
v0.11.8
v0.11.8
New Features
- Update LLM reasoning effort config and add Google Vertex AI support
- Add support for Qwen 3.5
- Improve side bar UI
- Automatically generate chat titles
Improvements & Bug Fixes
- Refactor model provider implementation
- Fix chat avatar display issues
v0.11.7
v0.11.7
Bug Fixes & Improvements
- Improved timeout error handling for more reliable agent responses during network issues
- Added environment and version information to alert notifications for better incident tracking
- Fixed CDP wallet skill functionality
- Improved error handling consistency across skill modules
- Fixed avatar display fallback in frontend activity and posts pages
- Fixed chat input disappearing on new thread in frontend
- Fixed tool call display in local UI
Infrastructure
- Added heartbeat monitoring to local Docker Compose setup
- Updated Go runtime to 1.26
v0.11.6
Bug Fixes
- Fixed an issue where "new conversation" in the agent chat interface would flicker or require multiple clicks.
Improvements
- Refactored the wallet interface to be fully asynchronous, improving performance and consistency across different wallet providers (Native, CDP, Safe).
Full Changelog: v0.11.5...v0.11.6
Release v0.11.5
New Features
- Added support for Langchain OpenRouter.
Bug Fixes
- Fixed an issue causing bugs in tests.
Improvements
- Improved exception handling for system skills.
- Updated project dependencies.
Full Changelog: v0.11.4...v0.11.5
Release v0.11.4
Release v0.11.4
Bug Fixes
- Fixed template model inheritance when creating agents.
- Fixed docker compose configuration issues.
Testing
- Switched to testing.postgres for database tests.
Improvements
- Updated project dependencies.
Full Changelog: v0.11.3...v0.11.4
v0.11.3
Release v0.11.3
New Features
- Added x402 updates to expand onchain payment support.
- Enabled frontend hot-reload in docker-compose for faster local development.
- Introduced LLM packer improvements to streamline model packaging.
- Launched the Hugo-based documentation site with initial installation guidance.
Improvements
- Improved avatar generation and frontend error handling for smoother UI experiences.
- Refined autonomous task defaults and display formatting.
- Updated image handling to use relative paths with CDN resolution.
- Stabilized local API behavior and development reload flows.
Full Changelog: v0.11.2...v0.11.3
v0.11.2
Release v0.11.2
New Features
- Added support for new AI models: MiniMax M2.5 with enhanced intelligence and structured output capabilities, and GLM 5 with improved performance and reasoning abilities.
Improvements
- Enhanced error logging in autonomous tasks with detailed exception information and stack traces for better debugging.
- Updated model configurations to reflect the latest available models and their capabilities.
Full Changelog: v0.11.1...v0.11.2
v0.11.1
Release v0.11.1
Improvements
- Introduced a unified base class for system skills, reducing code duplication and improving consistency across built-in agent capabilities.
- Added detection and error reporting for cases where the AI model produces an empty response, preventing silent failures during conversations.
Full Changelog: v0.11.0...v0.11.1