megatron-migration-generator
Pass
Audited by Gen Agent Trust Hub on Sep 20, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from an impact_report.json file, which is then rendered into markdown migration reports and checklists. If this JSON contains malicious instructions, they could influence the agent during subsequent review or application steps. \n
- Ingestion points: The impact_report.json file is read by render_migration_artifacts.py and synthesize_full_patch_series.py (referenced in SKILL.md). \n
- Boundary markers: The generated reports do not include delimiters or warnings to ignore instructions embedded within the report content. \n
- Capability inventory: The skill has the capability to edit local workspace files (apply mode) and create git commits (commit mode) as described in SKILL.md. \n
- Sanitization: While filenames are slugified to prevent path traversal, the text content within the generated reports and READMEs is interpolated directly from the JSON without escaping or validation. \n- [COMMAND_EXECUTION]: The skill executes local Python scripts (render_migration_artifacts.py and synthesize_full_patch_series.py) to transform JSON data into migration artifacts. These scripts are part of the skill's own package and perform legitimate file system operations within a user-specified output directory.
Audit Metadata