recovery-orchestrator
Pass
Audited by Gen Agent Trust Hub on Sep 10, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests data from external Zoho Billing reports which may contain untrusted strings such as customer names or notes.
- Ingestion points: Data is fetched via
ZohoBilling_get_lost_opportunities_reportandZohoBilling_get_abandoned_carts_reporttools. - Capability inventory: The agent writes the results to JSON and Markdown files and executes a Python script (
apply_recovery_matrix.py) to process the data. - Boundary markers: The instructions do not specify any delimiters or safety markers to isolate the report data from the agent's operational logic.
- Sanitization: No explicit sanitization or validation is mentioned for the content within the billing records before they are interpolated into the final proposal report.
- [COMMAND_EXECUTION]: The skill executes a local Python script to apply recovery logic to the fetched data.
- Evidence: Step 3 specifies the execution of
python <skill-path>/scripts/apply_recovery_matrix.pywith arguments pointing to workspace files.
Audit Metadata