lodestar-docs
Pass
Audited by Gen Agent Trust Hub on Nov 2, 2026
Risk Level: SAFECOMMAND_EXECUTIONDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The script
scripts/scope.mjsexecutes thegit logcommand vianode:child_process.spawnSyncto retrieve commit dates for files, which is used as a signal for the documentation audit. - [DYNAMIC_EXECUTION]: In
scripts/setup-modules.mjs, the skill uses dynamicimport()to load modules from a sibling directory (../../lodestar-setup/scripts/). This is a code-sharing mechanism between related skills from the same vendor to provide shared runtime and discovery logic. - [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted documentation content that could influence the agent's triage and deletion logic.
- Ingestion points: The agent is instructed to read the contents and headers of markdown files, including
README.mdandINDEX.md, to determine if they should be deleted, merged, or harvested (references/workflow.md). - Boundary markers: The instructions do not define boundary markers or include explicit warnings to the agent to disregard instructions found within the content of the audited documentation files.
- Capability inventory: The agent has the capability to delete files, rewrite documentation content, and commit these changes to the repository (
references/workflow.md). - Sanitization: Documentation content is processed directly by the agent for decision-making without a sanitization or filtering step to mitigate embedded instructions.
Audit Metadata