chonkie
Pass
Audited by Gen Agent Trust Hub on Jun 30, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill provides documentation and implementation examples for the 'chonkie' library, a legitimate high-performance text segmentation tool. All identified behaviors align with the stated purpose of building RAG ingestion pipelines.
- [EXTERNAL_DOWNLOADS]: The skill facilitates the installation of the 'chonkie' library and its optional dependencies from official package registries. It also references fetching configuration recipes from the 'chonkie-ai' organization on HuggingFace Hub, which is a standard distribution method for ML-related assets.
- [COMMAND_EXECUTION]: Provides standard CLI commands for environment setup (pip/uv installation) and starting a local development server (
chonkie serve). These commands are documented for user execution and do not involve obfuscated or automated execution of untrusted scripts. - [DATA_EXFILTRATION]: Network operations are limited to established third-party providers required for RAG functionality, including vector databases (Chroma, Qdrant, Pinecone, Weaviate, etc.) and embedding APIs (OpenAI, Cohere, Anthropic, VoyageAI). No unauthorized or suspicious data transmission patterns were detected.
Audit Metadata