add-stack-context
Pass
Audited by Gen Agent Trust Hub on Aug 5, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill is designed to identify a project's technology stack and fetch corresponding content (agent skills, documentation, or tools) from external vendor repositories and documentation sites.
- [COMMAND_EXECUTION]: The instruction to "install each vendor's official agent context... in the form the vendor recommends" implies the execution of shell commands to install tools, configure MCP servers, or apply project codemods.
- [PROMPT_INJECTION]: The skill facilitates the ingestion of third-party instructions ("agent context") into the project environment, creating an indirect prompt injection surface. Malicious or compromised vendor-provided instructions could be used to influence the agent's behavior in subsequent sessions.
- Ingestion points: External vendor documentation and public repositories identified at runtime (SKILL.md).
- Boundary markers: The skill does not specify the use of delimiters or boundary markers for the imported context.
- Capability inventory: The skill allows for file system modifications and the installation of new tools/skills into the workspace (SKILL.md).
- Sanitization: The instructions attempt to mitigate risk by mandating "Official sources only" and requiring the agent to "Skim what an install pulls in before accepting it."
Audit Metadata