Comprehensive guide for developing with IntentKit, an autonomous AI agent framework. Covers architecture, skills development, testing, and best practices for building agents with capabilities using LangGraph and LangChain.
Expert guidance for developing with IntentKit, an open and fair framework for building AI agents equipped with powerful skills.
This skill provides LLM agents with comprehensive knowledge of the IntentKit autonomous agent framework architecture, development patterns, and best practices. It enables agents to:
The intentkit package follows a strict import hierarchy to prevent circular dependencies:
**Import Order (left can import right, never reverse):**
`utils` → `config` → `models` → `abstracts` → `clients` → `skills` → `core`
1. **`intentkit/core/`** - Agent system powered by LangGraph
2. **`intentkit/models/`** - Entity models (Pydantic + SQLAlchemy)
3. **`intentkit/config/`** - System configuration (database, API keys)
4. **`intentkit/skills/`** - Skills system using LangChain BaseTool
5. **`intentkit/utils/`** - Utility functions (logging, formatting)
6. **`intentkit/abstracts/`** - Interfaces for core and skills
Before any development work:
```bash
source .venv/bin/activate
```
Use these specific tools and versions:
- ⚠️ Use only 2.0 API patterns, not legacy APIs
- ⚠️ Use only V2 interfaces, not obsolete V1
1. **Import Hierarchy:** Respect the package import order to avoid circular dependencies
2. **Import Placement:** Always place imports at the beginning of files
3. **Language:** Use English for all code comments
4. **Latest Packages:** Always use the latest package versions
5. **No Auto-Commit:** Do not commit changes unless explicitly requested
When developing or modifying agent skills:
1. Reference the detailed guide at `agent_docs/skill_development.md`
2. Skills are based on LangChain's `BaseTool`
3. Skills enable agents to fetch data, perform actions, and interact with environments
4. Follow the skills/ folder structure and conventions
Before finalizing any changes:
```bash
ruff format
ruff check --fix
pytest
```
- `tests/core/` - Core package tests
- `tests/api/` - API server tests
- `tests/skills/` - Skill integration tests
When performing Git operations (commits, pull requests, releases), reference the detailed guide at `agent_docs/ops_guide.md`.
1. **No Unnecessary Documentation:** Do not write dedicated documentation or example scripts after implementation unless explicitly requested
2. **English Only:** All code, comments, and searches must be in English
3. **Import Order:** Strictly follow the package import hierarchy
4. **Modern APIs Only:** Use SQLAlchemy 2.0 and Pydantic V2 patterns exclusively
5. **Type Safety:** Ensure BasedPyright validation passes
6. **Test Coverage:** Write and run tests for all changes
```python
from langchain.tools import BaseTool
from intentkit.models import MyModel # OK: skills can import models
from intentkit.config import get_api_key # OK: skills can import config
class NewSkill(BaseTool):
name = "new_skill"
description = "Description of what this skill does"
def _run(self, query: str) -> str:
# Implementation
pass
```
Leave a review
No reviews yet. Be the first to review this skill!
# Download SKILL.md from killerskills.ai/api/skills/intentkit-autonomous-agent-development/raw