atelier-distill
Pass
Audited by Gen Agent Trust Hub on Sep 28, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests and processes content from untrusted sources, including repository journals (.claude/LESSONS.md, .claude/PLAN.md) and the agent's internal memory folder.
- Ingestion points: Reads memory entries and files which could potentially contain malicious instructions embedded by external contributors to a repository.
- Boundary markers: The instructions include a specific safeguard directing the agent to treat processed entries as data and to ignore any instructions found within them.
- Capability inventory: The skill possesses file read/write capabilities and the ability to execute git commands through the agent's interface.
- Sanitization: Relies on explicit prompt-based instructions to prevent the execution of content found within the journals.
- [COMMAND_EXECUTION]: The skill invokes shell commands such as git show and generates git commit proposals to manage the versioning of memory files.
- These commands are transparently reported to the user and require explicit confirmation before execution, following standard agent safety protocols.
Audit Metadata