maintain-verification-skill
Pass
Audited by Gen Agent Trust Hub on Oct 12, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONPRIVILEGE_ESCALATION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests project source code and feature maps to detect documentation drift, creating a surface where malicious instructions in the analyzed files could influence the agent's behavior.
- Ingestion points: Project source code files (Step 2), the feature map README, and sibling files in the verification skill directory (Step 1).
- Boundary markers: Absent; the skill does not use specific delimiters or instructions to ignore embedded commands within the ingested source data.
- Capability inventory: File system modification within the verification skill directory, creation of symlinks, and execution of local harness scripts or launch commands.
- Sanitization: Absent; there is no evidence of sanitization or validation for the content read from the project source.
- [COMMAND_EXECUTION]: The skill is designed to run local "harness scripts" and drive applications using commands defined in the target verification skill's launch sections. This involves direct shell command execution on the host machine.
- [PRIVILEGE_ESCALATION]: In Step 0, the skill instructs the agent to perform file system operations including the creation of symlinks (
ln -s) to manage skill directory structures. While intended for path repair, manual symlink manipulation is a sensitive capability.
Audit Metadata