lokalise-common-errors

Pass

Audited by Gen Agent Trust Hub on Sep 9, 2026

Risk Level: SAFE
Full Analysis
  • [DATA_EXPOSURE_AND_EXFILTRATION]: The skill correctly uses the LOKALISE_API_TOKEN environment variable to handle authentication, which is a secure practice. Network operations via curl and urllib are directed solely to official Lokalise domains (api.lokalise.com, status.lokalise.com). No exfiltration patterns to third-party or unknown domains were found.
  • [REMOTE_CODE_EXECUTION]: No remote code downloads or execution from untrusted sources are present. The skill provides local diagnostic bash scripts and code examples that interact with the official Lokalise API.
  • [INDIRECT_PROMPT_INJECTION]:
  • Ingestion points: The skill processes PROJECT_ID, KEY_ID, and raw JSON responses from the Lokalise API (SKILL.md, implementation-guide.md).
  • Boundary markers: Not explicitly defined in code snippets, but instructions focus on structured data handling.
  • Capability inventory: Includes curl (network), bash (shell), and lokalise2 (CLI tool).
  • Sanitization: The skill provides logic for validating JSON formats and project ID structures, which serves as a mitigation for malformed or malicious API responses.
  • [OBFUSCATION]: No obfuscated strings, Base64 commands, or hidden Unicode characters were detected in the instructions or code references.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 9, 2026, 03:44 AM
Security Audit — agent-trust-hub — lokalise-common-errors