Self-Improving AI Agents: How Alice Learns Skills From Experience
Alice Agent creates and improves skills automatically after complex tasks. Heres how the self-improvement loop works.
Most AI agents are stateless — they forget everything between conversations. Alice Agent has a built-in learning loop that makes it genuinely self-improving.
The Learning Loop
After completing a complex task, Alice can automatically create a skill — a reusable chunk of procedural memory. Skills capture patterns: how to deploy to a specific platform, how to format code reviews, how to interact with an API.
# Alice creates skills automatically, or you can install them:
alice skills install github-code-review
alice skills install starrocks-data-explorationHow Skills Improve
Skills are not static. When Alice uses a skill and encounters a better approach, it patches the skill in place. This means the skill gets more effective every time it is used — a genuine improvement loop.
Memory Across Sessions
Alice stores facts, preferences, and project context in persistent memory. When you start a new session, Alice already knows your preferred coding style, project architecture decisions, previous conversations and outcomes, and custom configurations.
The Curator
A background process tracks skill usage and auto-archives stale ones. You never lose skills — archives go to ~/.alice/skills/.archive/ and are restorable. Pinned skills are protected from deletion.
Why This Matters
A self-improving agent gets more valuable over time. Instead of re-explaining context every session, Alice builds on what it already knows. The more you use it, the better it gets.