project-add-items
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill constructs shell commands using the
ghCLI by interpolating content (titles and descriptions) parsed from external files and URLs. This presents a potential command injection risk if the source material contains shell metacharacters like backticks or semicolons. - [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest data from untrusted sources to influence its behavior, specifically for creating and editing GitHub Project items.
- Ingestion points: Markdown/YAML files, external URLs (GitHub Wiki, Issue lists), and user-provided inline text (Step 1).
- Boundary markers: None. The instructions do not specify the use of delimiters or warnings to ignore embedded instructions in the source text.
- Capability inventory: Execution of
gh projectcommands includingitem-createanditem-edit(Steps 4 and 5). - Sanitization: None. The skill does not describe any validation or escaping of the parsed content before it is used in CLI commands.
Audit Metadata