frame-ml-problem
Pass
Audited by Gen Agent Trust Hub on Oct 15, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes various subcommands of the
skore-skillsCLI tool usingpython -m skore_skills. This includes commands for viewing status, clearing specific journal entries based on user input, building a documentation site, and managing git turns. - [INDIRECT_PROMPT_INJECTION]: The skill incorporates an attack surface for indirect prompt injection by processing external documentation.
- Ingestion points: The skill is instructed to read EDA reports, free-form notes, and README files associated with datasets to extract facts for modeling decisions in
SKILL.md(Step 5). - Boundary markers: There are no explicit instructions to use delimiters or ignore embedded commands when the agent reads these external files.
- Capability inventory: The skill has the ability to execute shell commands through the
skore-skillspackage, write tojournal/JOURNAL.md, and load other skills (e.g.,setup-ml-project,model-ml-pipeline). - Sanitization: The skill relies on matching extracted information against predefined candidate lists for certain fields, but does not specify sanitization for free-text fields or general file reading.
Audit Metadata