add-locale
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill instructions allow for user-provided input to be interpolated directly into shell commands and code files without specifying validation or sanitization rules.
- Ingestion points: User-provided variables
${isoAlpha2}and${localeName}are requested at the start of the skill execution. - Boundary markers: Absent. The skill does not provide any instructions to the agent to verify that the input conforms to expected ISO 639-1 or BCP-47 patterns (e.g., regex checks).
- Capability inventory: The skill leverages
Bash(forgitandpnpmoperations) andWrite/Edit(for source code modification). - Sanitization: Absent. The user input is directly embedded into branch names, commit messages, filenames, and source code identifiers.
- [COMMAND_EXECUTION]: The skill uses user-supplied variables within shell command strings (e.g.,
git checkout -b add-locale-${isoAlpha2}andgit commit -m "Register ${isoAlpha2}"). If the agent executes these commands without internal escaping, it presents a potential command injection vector.
Audit Metadata