skills-repo-maintenance
Pass
Audited by Gen Agent Trust Hub on Sep 29, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands such as
git rev-parseto identify the repository root andpythonto run a local packaging script. - The packaging scripts are located within the local environment path (
$HOME/.codex/skills/.system/skill-creator/scripts/package_skill.py), which is consistent with the maintenance workflow of the Codex platform. - [INDIRECT_PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface as it ingests and processes skill metadata and repository structures to update configuration files.
- Ingestion points: Reads
SKILL.mdfiles and repository directory structures (SKILL.md). - Boundary markers: None explicitly defined to delimit external content from instructions.
- Capability inventory: Includes the ability to execute Git commands and run local Python scripts for file packaging (SKILL.md).
- Sanitization: The instructions do not specify validation or sanitization of the metadata content retrieved from the repository files before it is processed or included in marketplace manifests.
Audit Metadata