The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill retrieves and processes untrusted data from ServiceNow tables (such as case descriptions, work notes, and emails) to generate resolution documentation. This creates a surface where malicious instructions embedded in the case data could potentially influence the agent's behavior during note generation.
Ingestion points: Data fetched via ServiceNow MCP tools and REST endpoints (e.g., sn_customerservice_case, sys_journal_field, sys_email) in Steps 1 through 7.
Boundary markers: The template in Step 8 uses headers like === RESOLUTION NOTES ===, but lacks specific instructions to ignore instructions or commands embedded in the source data.
Capability inventory: The skill uses Bash and ServiceNow tools, including the ability to write to resolution_notes and close_notes fields.
Sanitization: There are no documented steps to filter, escape, or sanitize the retrieved content before interpolation into the final output template.

resolution-notes