zcode-cli
Pass
Audited by Gen Agent Trust Hub on Oct 16, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill involves the execution of the Zcode CLI or a bundled Node.js script (zcode.cjs) to perform tasks. It provides specific instructions for locating these local rountimes and executing them with various flags.\n- [INDIRECT_PROMPT_INJECTION]: The skill is designed to process external content from the workspace (such as README files and source code), which creates a surface for indirect prompt injection.\n
- Ingestion points: The skill reads source code and documentation from the local directory defined by the --cwd flag.\n
- Boundary markers: The instructions do not specify the use of explicit text-based delimiters, but they rely on structured command-line parameters for prompt isolation.\n
- Capability inventory: The skill can execute subprocesses and interact with the file system via the Zcode CLI interface.\n
- Sanitization: The risk is mitigated by recommending explicit permission modes (e.g., --mode plan) and tool exclusions (--disallowed-tools) to restrict the potential impact of instructions embedded in external data.
Audit Metadata