migrate-airflow-kestra
Pass
Audited by Gen Agent Trust Hub on Sep 17, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill downloads a JSON schema from api.kestra.io to ensure generated flows are production-ready and schema-validated. This is a trusted vendor resource for this skill.
- [DYNAMIC_EXECUTION]: A Python one-liner is used to extract specific task definitions from the downloaded JSON schema. The script performs string searching and printing; it does not execute the data contained within the downloaded file as code.
- [COMMAND_EXECUTION]: The skill uses shell commands to read the source DAG file, perform schema fetching, and write the resulting Kestra YAML and Python namespace files to the local file system.
- [INDIRECT_PROMPT_INJECTION]: The skill ingests user-provided Airflow DAG files for processing. While this represents a vulnerability surface for indirect instructions, the skill defines strict mapping rules and emphasizes the use of secure secret management (Pebble expressions) rather than hardcoded credentials. Evidence Chain: 1. Ingestion: Airflow DAG .py file via $ARGUMENTS. 2. Boundary markers: None. 3. Capability inventory: curl, Python execution, file system read/write. 4. Sanitization: Not explicitly implemented in script, relies on structural mapping logic.
Audit Metadata