create-pr
Pass
Audited by Gen Agent Trust Hub on May 2, 2026
Risk Level: SAFE
Full Analysis
- [PROMPT_INJECTION]: The skill has an indirect prompt injection surface as it ingests git branch names, commit messages, and repository diffs to generate PR titles, summaries, and playground descriptions.
- Ingestion points: Repository metadata, commit history, and code diffs are read during the context gathering phase in
SKILL.md. - Boundary markers: The skill employs HEREDOC delimiters (
EOF) when passing generated content to theghCLI, which provides some isolation between the untrusted data and the shell command. - Capability inventory: The skill uses
Bashfor git operations and theSkilltool to invoke external plugins. - Sanitization: Basic sanitization is performed on branch names (replacing slashes with double hyphens) before they are used in file system paths.
- [EXTERNAL_DOWNLOADS]: The skill references and recommends the installation of the
playgroundplugin from Anthropic's official GitHub organization (anthropics/claude-plugins-official). This is a trusted source used for generating interactive PR previews. - [COMMAND_EXECUTION]: The skill uses the
Bashtool to execute standard development commands includinggit,gh(GitHub CLI), and project-specific linters/test runners. These operations are restricted to the local repository context and are necessary for the skill's primary function of PR automation.
Audit Metadata