agentic-context-engine
🧠Make your agents learn from experience. Now available as a hosted solution at kayba.ai
Agentic Context Engine (ACE) is an agent skill framework that adds persistent learning capabilities to AI agents through experience-based strategy accumulation. Instead of repeating the same mistakes across sessions, ACE maintains a Skillbook—a collection of strategies that evolves as the agent executes tasks and receives feedback. The framework uses three specialized roles (Agent, Reflector, and SkillManager) to analyze execution traces, extract actionable insights, and continuously refine strategies. Built on PydanticAI, it supports 100+ LLM providers and integrates with popular tools like browser-use, LangChain, and Claude Code.
Key Features
Use Cases
- 01Building AI agents that improve accuracy over time by learning from past mistakes and successes
- 02Browser automation tasks that become more efficient as the agent discovers optimal strategies
- 03Multi-step agentic workflows requiring tool use and policy adherence with consistent execution
- 04Code translation and autonomous development tasks with minimal supervision
- 05Extracting reusable strategies from historical agent logs and execution traces
- 06Production agents that need to reduce token costs through learned optimization patterns
Related Skills
View moresuperpowers
An agentic skills framework & software development methodology that works.
skills
Skills for Real Engineers. Straight from my .agents directory.
skills
Public repository for Agent Skills
ponytail
Makes your AI agent think like the laziest senior dev in the room. The best code is the code you never wrote.
agentic-context-engine — FAQ
What is Agentic Context Engine (ACE)?+
ACE is an open-source framework that adds persistent learning capabilities to AI agents by maintaining a Skillbook of strategies extracted from execution traces. It enables agents to learn from experience and avoid repeating mistakes across sessions.
How do I install ACE?+
Install ACE using uv with 'uv add ace-framework'. Run 'ace setup' for interactive configuration, or manually set your API key (e.g., export OPENAI_API_KEY='your-key'). Additional runners like browser-use or LangChain require optional dependencies: 'uv add ace-framework[browser-use]'.
Which LLM providers does ACE support?+
ACE supports 100+ LLM providers through its PydanticAI and LiteLLM integration, including OpenAI, Anthropic, Google, AWS Bedrock, and Groq. Any LiteLLM model string can be passed to ACELiteLLM.
Do I need API keys to use ACE?+
Yes, you need an API key for your chosen LLM provider (e.g., OPENAI_API_KEY or ANTHROPIC_API_KEY). The 'ace setup' command walks you through configuration and connection validation.
Is ACE free to use?+
ACE is open-source and free to use. You only pay for LLM API calls to your chosen provider. A hosted managed service called Kayba is available separately at kayba.ai.
Can I use ACE with existing agent frameworks?+
Yes, ACE provides integrations for LangChain, browser-use, and Claude Code through specialized runners. You can also analyze existing execution traces with TraceAnalyser without re-running tasks.
How do I install agentic-context-engine?+
Open the source repository on GitHub and follow its README. agentic-context-engine is a skill — MCP Agents Market links you directly to the official repo.
Is agentic-context-engine free?+
agentic-context-engine is an open-source project hosted on GitHub. Check the repository for its license and any usage requirements.