adaptive-learning-agents
от vedantsingh60
Personal DevelopmentКратко
**Learn from errors and corrections in real-time.
Описание
Что делает этот навык
A Python library that records errors, corrections, and successful patterns locally, then retrieves relevant entries before tackling similar problems. All data stays on disk with no external dependencies, API keys, or network calls required.
Because it requires zero dependencies and no external accounts, it works immediately in any Python environment without setup friction.
Когда его использовать
- Logging a bug fix so it surfaces the next time the same error appears
- Tracking which prompt formats produce reliable outputs for a given task
- Documenting API quirks and workarounds discovered during integration work
- Exporting accumulated knowledge as JSON to share across a team or project
- Reviewing unresolved errors before starting a complex or high-stakes task
Пример сценария
User asks: Record that the Anthropic API rejects prompts with excessive newlines and note the fix
- 1
Call record_error with the error_description, context, and solution fields populated
- 2
Store the entry locally in the .adaptive_learning/ directory
- 3
Later, call search_learnings('newlines') to retrieve the stored entry
- 4
Return the matching learning with the solution and any prevention tip
The stored error entry is returned with the solution: use strip() to clean whitespace before sending prompts to the API
Установка
clawhub install vedantsingh60/adaptive-learning-agentsВыполните команду CLI или вставьте промпт в Claude Code, Cursor, Codex — любой кодинг-агент, которым пользуетесь.