tg-rich-digest
Pass
Audited by Gen Agent Trust Hub on Aug 8, 2026
Risk Level: SAFEPROMPT_INJECTIONEXTERNAL_DOWNLOADSDATA_EXFILTRATION
Full Analysis
- [PROMPT_INJECTION]: As the skill is designed to ingest and summarize external data (e.g., community reports, newsletters, status updates), it possesses an indirect prompt injection surface where malicious inputs could attempt to manipulate the agent's summary output. Evidence: The skill defines ingestion points for various community reports in SKILL.md. Boundary markers for untrusted content are not explicitly included in the provided templates. The skill's capability includes network operations via the Telegram Bot API. To mitigate these risks, the skill includes a 'Preflight gate' (SKILL.md) that instructs on the verification of links and formatting before the message is published.
- [EXTERNAL_DOWNLOADS]: The skill provides templates that reference external media assets from example.com. This is standard for document-grade rich messages and uses common documentation placeholders.
- [DATA_EXFILTRATION]: The skill provides instructions and examples (SKILL.md, community-digest.json) for sending data to the Telegram Bot API (api.telegram.org). This is the intended purpose of the skill and involves communication with a well-known service.
Audit Metadata