v0.2.2

Key Changes

• Agent Self-Scheduling (Part 1): Transitioned from a fixed, predefined interval to a dynamic, agent-driven scheduling system. This update lets agents determine the next workflow step based on real-time instructions. Part 2 will include a robust task scheduling system.
• Web Search Tool Integration: Implemented a web search tool to enhance information retrieval capabilities, providing agents with stronger research capabilities
• Autonomous Twitter Agent Workflow: Launched a new workflow that enables the Twitter agent to operate independently, improving agent autonomy.
• Experience Minting on Autonomys Network Mainnet: Added the capability to mint experiences directly on the Autonomys Network mainnet via Auto-Drive
• Enhanced Agent Composition Examples: Provided multiple practical examples to demonstrate how to compose and configure agents, making it easier for users to get started.
• Improved Configuration Management: Introduced multiple enhancements to configuration management, resulting in a more robust and flexible setup process.

What's Changed

• Refactor orchestrator workflow to be instantiable with tools and prompts by @jfrank-summit in #193
• Summarization Feature in Orchestrator Workflow by @Xm0onh in #192
• Twitter tool cleanups by @jfrank-summit in #194
• Agent Memory Viewer: Refactored Agent Watcher & YAML Schema Updates by @Xm0onh in #195
• Support o3-mini as samll model for reasoning by @Xm0onh in #196
• Refactor LLM Configuration: Replace LLMSize with provider & model by @Xm0onh in #202
• Upgrade Auto-drive package and add Network to config file by @Xm0onh in #203
• Add autonomous twitter agent tool by @jfrank-summit in #204
• Allow for customizable prompt templates by @jfrank-summit in #208
• Integrating Vector DB with orchestration workflow and nodes by @Xm0onh in #210
• Move workflow control to state to simplify parsing by @jfrank-summit in #211
• Only bind tools to input node by @jfrank-summit in #213
• Refactor Orchestrator Configuration for Instance-Level Model Selection and Context Management by @Xm0onh in #215
• Feature: Agent can self-schedule workflow by @jfrank-summit in #217
• Add Autonomous and Directed Twitter Agent Examples by @jfrank-summit in #218
• Add webSearchTool by @jfrank-summit in #219

Full Changelog: v0.2.1...v0.2.2

v0.2.1

What's Changed

• Enhancing resurrection - Clean coding by @Xm0onh in #173
• Create version of twitter tools more optimized for orchestration workflow by @jfrank-summit in #174
• Update README.md to include VS C++ redistributable dependency for windows users by @jfrank-summit in #175
• Update package.json build command for Memory-Viewer by @Xm0onh in #176
• Quick fix for inconsistency in signature by @Xm0onh in #178
• Read agent version from package.json by @Xm0onh in #179
• Prompt optimizations - Improve ability to autonomously carry out workflows by @jfrank-summit in #180
• Add tools to like tweets, follow users and fetch the "following" timeline by @jfrank-summit in #181
• Return useful info from twitter workflow instead of final state by @jfrank-summit in #184
• Add fetchFollowing and fetchProfile tools by @jfrank-summit in #190
• Add tool for retrieving current time by @jfrank-summit in #191

Full Changelog: v0.2.0...v0.2.1

v0.2.0

PLEASE NOTE: This repo is in very active development with breaking changes expected frequently.

Key Changes

• Added orchestration workflow which moves from pre-defined workflows to more autonomous planning and exectution NOTE: This workflow is not yet optimized for complex workflows. Improvements to token usage will be coming shortly. Currently, only Claude Sonnet 3.5 reliably executes complex workflows as intended. The previous, pre-defined Twitter workflow can be executed by passing "Run Twitter workflow" as the initial input.
• Added multi-character setup
• Added initial support for vector database (vectorlite)
• Added saveExperience tool for saving unstructured experiences or memories permanently to the Autonomys Network

All Changes

• Feat/memory viewer timeline by @Xm0onh in #145
• Update README.md - clarify necessary env variables for Auto Drive integration by @jfrank-summit in #146
• fix auto_drive typo in config.example.yaml by @jfrank-summit in #149
• Fix character copy command by @clostao in #148
• Enhance Character Selection UX and Fix Minor Typos by @Xm0onh in #153
• Change KOL workflow to Twitter workflow by @jfrank-summit in #156
• Feature - Add Vector Database and Resurrection Memory Script by @Xm0onh in #152
• Update fav icon of memory-viewer for argumint by @Xm0onh in #157
• Fix a minor typo in README by @Xm0onh in #159
• Local Memory Hash Validation by @Xm0onh in #158
• Add Orchestration Workflow by @jfrank-summit in #160
• Make config multi-character friendly by @jfrank-summit in #161
• Fix thread issue on memory viewer by @Xm0onh in #164
• Add Autonomys Agents banner to README.md by @jfrank-summit in #163
• Refactor Character Folders by @jfrank-summit in #162
• Allow for workspace wide envs and specific character envs by @marc-aurele-besner in #165
• Enhanced Orchestration Process - Autonomous Planning and Execution by @Xm0onh in #166
• Add a second login attempt if using the cookie fail by @marc-aurele-besner in #168
• Add data related tools to orchestrator workflow (vectorDb and saveExperience) by @jfrank-summit in #169
• Refactor Agent Memory Viewer - Supporting Generalized Types and multiple agents by @Xm0onh in #170
• Add deepseek model support by @jfrank-summit in #171

New Contributors

• @clostao made their first contribution in #148

Full Changelog: v0.1.0...v0.2.0

v0.1.0

Initial implementation of Autonomys Agents framework for building fully autonomous and onchain agents.

What's Changed

• Add zod for config format - clean coding and refactoring by @Xm0onh in #97
• Improve Twitter client interactions by @jfrank-summit in #100
• Add cli option to select a character file by @jfrank-summit in #103
• Feat/agent frame summary by @Xm0onh in #101
• Memory viewer prod - Enhance security and add search option by @Xm0onh in #104
• Ref/agent frame types by @Xm0onh in #106
• Bug/agent frame latest cid by @Xm0onh in #107
• Bug/agent frame skipped response date by @Xm0onh in #109
• Update post tweet logic and other minor fixes by @jfrank-summit in #110
• Feat/agent frame mul llm by @Xm0onh in #112
• Bug/circular serializing tweets by @Xm0onh in #115
• Ref/separate timeline response by @Xm0onh in #116
• fix issue #113 by @Xm0onh in #121
• refactoring .env vars - remove unconfidential vars by @Xm0onh in #119
• Feat/af get cid local by @Xm0onh in #122
• Feat/af test scripts by @Xm0onh in #125
• Ref/af following typo by @Xm0onh in #126
• fix processedTweetIds bug issue #129 by @Xm0onh in #130
• Feat/af memory pruning by @Xm0onh in #128
• Tweet Processing Enhancements by @jfrank-summit in #134
• Feat/af lint by @Xm0onh in #137
• Ref/lintfix by @Xm0onh in #138
• refactoring the repo by @Xm0onh in #139
• Refactor configuration (character and Auto Drive upload flag) by @jfrank-summit in #142
• Feat/ds version memory viewer by @Xm0onh in #143
• Update project description in README.md by @jfrank-summit in #144
• Add build ci/cd workflow by @marc-aurele-besner in #141

Full Changelog: https://github.com/autonomys/autonomys-agents/commits/v0.1.0