init-vscode
Pass
Audited by Gen Agent Trust Hub on Sep 29, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONDYNAMIC_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill reads and merges existing project configuration files, which presents a surface where malicious instructions embedded in these files could potentially influence agent behavior.
- Ingestion points: Reads existing configuration files including .vscode/settings.json, .vscode/extensions.json, package.json, and Prettier configuration files from the project root.
- Boundary markers: Absent. The skill does not use specific delimiters or instructions to ignore embedded instructions in the ingested data.
- Capability inventory: The skill can write to the filesystem, create directories via mkdir, and execute shell commands via pnpm exec.
- Sanitization: Absent. The skill relies on JSONC parsing and validation but does not include sanitization against natural language instructions.
- [COMMAND_EXECUTION]: The skill executes shell commands for environment setup and tool validation.
- Evidence: Uses mkdir -p .vscode to ensure the configuration directory exists within the project.
- Evidence: Uses pnpm exec prettier to validate the formatting and syntax of generated JSONC configuration files.
- [DYNAMIC_EXECUTION]: The skill dynamically modifies project configuration files by appending JavaScript code.
- Evidence: The skill logic includes steps to append or modify the overrides section in prettier.config.mjs to ensure .vscode/extensions.json is correctly handled by the JSONC parser.
Audit Metadata