yandex-direct-client-lifecycle
Pass
Audited by Gen Agent Trust Hub on May 11, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill uses official Yandex Search APIs and the Firecrawl scraping service to gather competitive intelligence. These operations are core to the skill's purpose and use authenticated endpoints (https://searchapi.api.cloud.yandex.net and https://api.firecrawl.dev).
- [COMMAND_EXECUTION]: The skill includes several utility scripts for scaffolding projects, processing TSV files, and capturing search results. These scripts utilize standard libraries (requests, lxml, playwright) to perform data aggregation and reporting.
- [DATA_EXPOSURE_AND_EXFILTRATION]: No evidence of exfiltration of sensitive user data was found. The skill encourages best practices for secret management by using environment variables (YANDEX_SEARCH_API_KEY, FIRECRAWL_API_KEY) or local configuration files located outside of version control.
- [OBFUSCATION]: The skill contains no malicious obfuscation. While it utilizes Base64 decoding, this is specifically implemented to handle the 'rawData' field returned by the official Yandex Search API, which is the documented method for retrieving HTML/XML response content. The homoglyph detections are false positives related to legitimate Russian language content and project-specific identifiers.
- [INDIRECT_PROMPT_INJECTION]: The skill has an ingestion surface for untrusted data as it scrapes external web pages and sitemaps. However, the workflow includes mandatory 'human-review' steps and strictly structured data processing (TSV/Markdown), which significantly mitigates the risk of the agent following instructions embedded in scraped content.
- [SAFE]: The 'build_secure_client_report.py' script generates a self-contained HTML report with SHA-256 password protection for client-facing deliverables. This is a legitimate security feature for protecting business research data.
Audit Metadata