manage-ml-backlog
Pass
Audited by Gen Agent Trust Hub on Oct 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill reads and processes data from external files like journal/ideas/*.md and audit digests, then incorporates this content into JOURNAL.md or design notes. Ingestion points: idea files, audit digests, and report.txt files. Boundary markers: No explicit delimiters or warnings are used when interpolating untrusted data. Capability inventory: The skill can read and write files, execute git operations, and call the skore_skills CLI. Sanitization: No specific sanitization or escaping is mentioned; the skill often pastes content verbatim or summarizes it.
- [COMMAND_EXECUTION]: The skill frequently executes shell commands via the
python -m skore_skillsCLI to manage workspace state, scaffolding, and git synchronization.
Audit Metadata