lantern lint
lantern lint is AgentLantern's static checker for agent projects.
It catches design issues before runtime: missing agents, invalid task references, incomplete YAML fields, suspicious delegation settings, wrong field types, and other project structure problems.
It does not call an LLM and does not require network access.
Pages
| Page | Use it for |
|---|---|
| Run the Linter | Basic commands, strict mode, JSON mode |
| CrewAI Rules | Error, warning, and info rule reference |
| JSON and CI | Machine-readable output, exit codes, GitHub Actions, pre-commit |
| Troubleshooting | Common linter problems |
How It Works
The linter reads the same project model that lantern docs uses:
- YAML config files,
- Python AST,
- framework-specific analyzer output.
Then it applies rules specific to the detected framework.
Current Coverage
| Framework | Linter status |
|---|---|
| CrewAI | Full rule set |
| Google ADK | Framework rules |
| LangGraph | Graph rules |
| AutoGen | Common Python-agent rules |
| Smolagents | Common Python-agent rules |
| OpenAI Agents SDK | Common Python-agent rules |
| Additional Python ecosystems | Common Python-agent rules |
| JavaScript/TypeScript ecosystems | Common JS-agent rules |
| Semantic Kernel .NET/Java | Semantic Kernel rules |
| Other Python or JS/TS frameworks | Common agent rules |