Skip to content

Changelog

[1.1.10-beta.2] - 2024-11-21

Summary of Changes

The pull request introduces several updates and improvements across multiple files in the project:

  • Template Renaming: The gm template has been renamed to gpt throughout the codebase. This reflects a shift to using GPT-based agents for more advanced interactions.

  • Prompt Enhancements: Agent prompt functions have been updated to use a simplified and more effective prompt generation method. The prompts now leverage a defaultPromptTemplate, improving the way agents generate and handle prompts for GPT-based models.

  • Skill Definitions Updates: Skill definitions in skills.ts files have been modified. The triggers property has been removed from skill objects, and parameters have been updated to enhance skill parsing and execution.

  • New Test Suites: New test files have been added to improve testing coverage for agents and skill parsing. These include tests for agent prompting, skill parsing, and group interactions.

  • Codebase Cleanup: Outdated files and templates, such as the old gm template and middleware files, have been removed to streamline the codebase and reduce clutter.

  • Documentation Updates: Documentation has been extensively updated to reflect all changes. This includes updates to READMEs, code examples, tutorial guides, and the AI Skills framework documentation.

  • Dependency and Configuration Updates: Several dependencies have been updated to newer versions, and configurations like package versions and build scripts have been adjusted accordingly.

  • Improvements in Handlers and Skills: Handler functions have been refined for better error handling and functionality, particularly in agent and group templates.

  • User Information Resolution: Enhancements have been made to how user information is resolved and cached, improving efficiency and consistency when retrieving user data like ENS domains and addresses.

Overall, these changes aim to enhance the functionality of GPT-based agents, improve code quality, update documentation for better developer experience, and streamline the project structure.

[1.1.8] - 2024-11-11

Summary of Changes

  • Introduced new commands for handling group interactions, including /id for retrieving group ID.

  • Added a new section for "AI Skills" in the documentation, detailing a framework for building AI agents.

  • Enhanced command execution with improved error handling and response structures.

  • Streamlined command handling for gaming interactions and tipping features.

  • Expanded example responses in the agent_prompt function for ENS domain interactions.

  • Added a new handler for group commands to manage "help" and "id" requests.

  • Modified the /exists command to allow access for all users, removing the admin-only restriction.

  • Bug Fixes
    • Resolved issues related to command execution and improved logging for better error tracking.
  • Documentation
    • Updated installation instructions and added more resources for users.
    • Enhanced clarity in documentation regarding command processing and multi-line responses.
    • Modified the processing documentation to reflect changes in handling multi-line responses.
    • Removed references to STACK_API_KEY in documentation, indicating it is no longer required.

  • Chores - Removed outdated files and consolidated functionality to improve code organization. - Simplified development setup by removing unnecessary dependencies and replacing nodemon with node --watch. - Introduced post-install scripts for TypeScript compilation.

[1.1.5] - 2024-11-02

Summary of Changes

-Tagline: An open framework for building AI agents ? +Tagline: An open framework for building AI agents

  • Lighter but more structured and robust.
  • Updated directory
  • Each agent can have its own landing in messagekit (devs can do PRs)
  • Quickstarts section
    • Skills: There is a 4-step approach to AI Skills:
    • Definition: Define the skills
    • Parsing: From @base please let swap 1 eth to usdc to /swap 1 eth usdc
    • Processing: For an agent to process new lines
    • Scenarios: For the agent to understand the context and perform skills autonomously.
  • Reasoning: For the agent to understand all context and perform skills autonomously.
  • Resolver: Fetches context from the user like ENS domain or Converse username

MessageKit updates

Summary of Changes Documentation:
  • Restructured directory
  • Added quickstarts.
  • Replaced one-to-one with Agent example
  • Documented skills
  • Renamed examples as templates
  • How to use a second xmtpClient client
  • Optional parameters concepts/structure#optional-parameters
  • Added changelog

Middleware: Added Middleware section as the layer where to interact with external services in general.

  • GPT: Use gpt responses (Updated)
  • Cron: Send updates daily
  • Resolver: Resolve ens and converse usernames
  • Notion: Interact with notion docs
  • Redis: Interact with Redis DB
  • Stackso: Use stackso for tracking points and loyalty programs
  • Vision: Use GPT vision to parse attachments
  • Learnweb3: Use learnweb3 libraries for testnet funds
Bug Fixes:
  • Fixed CLI issues when install with npm npm init message-kit