vscode
Pass
Audited by Gen Agent Trust Hub on Mar 1, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill provides bash command templates using the
codeandgitCLI tools to display diffs. It uses a fixed temporary file path/tmp/oldto store intermediate content, which is a poor practice in multi-user environments. - [PROMPT_INJECTION]: The skill creates an indirect prompt injection surface by recommending shell commands that interpolate potentially untrusted user data.
- Ingestion points: File paths and git references provided as arguments to the
codeandgitcommands (SKILL.md). - Boundary markers: No markers or escape sequences are used to isolate these inputs from the shell command structure.
- Capability inventory: Execution of arbitrary local commands via the shell with the capabilities of the
gitandcodebinaries (SKILL.md). - Sanitization: No sanitization or validation of the provided paths or commit hashes is defined within the skill.
Audit Metadata