mlflow-python

Pass

Audited by Gen Agent Trust Hub on Nov 7, 2026

Risk Level: SAFEPROMPT_INJECTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The SKILL.md file contains a 'Self-Evolving Skill' section that instructs the agent to modify the skill's own instructions based on its usage experiences. This directive for the agent to rewrite its own logic can be exploited to bypass intended constraints or introduce persistent malicious behavior if the agent is influenced by adversarial input during a session.- [INDIRECT_PROMPT_INJECTION]: The skill includes scripts that ingest untrusted data from external sources, creating a potential attack surface.
  • Ingestion points: The scripts/log_backtest.py script reads user-provided CSV files through the --returns argument.
  • Boundary markers: The instructions lack explicit delimiters or guidance to treat the imported CSV content as purely data, increasing the risk that embedded strings could be misinterpreted as instructions by the agent.
  • Capability inventory: The skill is permitted to perform file system operations (Read, Bash, Grep, Glob) and network communications (via the MLflow API).
  • Sanitization: The script relies on standard pandas.read_csv parsing without specific sanitization of headers or data values, which could lead to injection if the resulting logged metrics or parameters are subsequently interpreted by an LLM.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 7, 2026, 02:12 AM