zj-agents-init
Pass
Audited by Gen Agent Trust Hub on Sep 4, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill directs the agent to explore the repository environment using common shell commands such as
git remote -vand checks for existing files like.git/configandCLAUDE.md. It also provides operational patterns for the official GitHub (gh) and GitLab (glab) CLI tools to manage issues and pull requests. - [INDIRECT_PROMPT_INJECTION]: The skill possesses an indirect injection surface because it reads and processes existing repository content (untrusted data) to determine how to scaffold the new configuration.
- Ingestion points: Reads contents of
.git/config,AGENTS.md,CLAUDE.md, and scans the directory structure (e.g.,docs/,src/). - Boundary markers: Uses standard Markdown sectioning and documentation headers in the seeded templates.
- Capability inventory: Performs file reads, file writes to the local repository (
docs/zj-agents/,AGENTS.md,CLAUDE.md), and executes repository discovery commands. - Sanitization: The instructions do not specify any sanitization or escaping of the ingested repository metadata before using it to drive the setup logic.
Audit Metadata