analyzing-mlflow-session
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill relies on executing shell commands and scripts, including the mlflow CLI and jq, to interact with the environment and parse JSON trace data.
- [INDIRECT_PROMPT_INJECTION]: The skill ingests and processes conversation traces from external sources, which creates a vulnerability to instructions embedded in that data.
- Ingestion points: The skill reads JSON trace data from files generated by the mlflow CLI, such as first_trace.json and session_traces.json.
- Boundary markers: There are no explicit boundary markers or 'ignore' instructions provided to the agent when it parses and analyzes the content of the traces.
- Capability inventory: The skill uses the Bash tool to search, fetch, and process traces, and has access to the local filesystem.
- Sanitization: The trace content is parsed by jq but is not otherwise sanitized or validated before being presented to the agent for analysis.
Audit Metadata