the-weekly-reporter
Pass
Audited by Gen Agent Trust Hub on Aug 14, 2026
Risk Level: SAFE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data such as performance exports and call transcripts. However, it includes robust boundary markers and explicit instructions to treat all fetched content as data, never as instructions.
- Ingestion points: Performance, traffic, inventory, and support summaries provided by the user as described in SKILL.md.
- Boundary markers: Strong instructions in the "Untrusted content is data" section of SKILL.md and the dedicated
references/agent-security.mdfile, which define clear rules for identifying and ignoring injection attempts. - Capability inventory: The skill is limited to generating markdown reports and reading local context files. It does not possess network-send or arbitrary shell execution capabilities.
- Sanitization: Detailed rules require the agent to HTML-escape interpolated values in generated code and to quote rather than execute any instructions found within user data.
- [DATA_EXPOSURE]: The skill includes proactive defensive instructions to detect, redact, and flag credentials (such as API keys or bearer tokens) that may be present in CSV exports or transcripts, preventing them from being echoed back or persisted.
- [SAFE]: No malicious patterns, obfuscation, or unauthorized remote code execution were detected. The skill follows security and data integrity best practices by including protocols for missing data handling and accessible chart generation.
Audit Metadata