The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests and processes untrusted data from ServiceNow HR cases, creating a surface for potential indirect prompt injection attacks.
Ingestion points: The skill retrieves content from fields such as description in the sn_hr_core_case table and work_notes_list in the sn_hr_core_task table (SKILL.md, Steps 1 and 4).
Boundary markers: The summary procedure uses markdown headers and visual delimiters (e.g., ---) to structure the output, but it does not include explicit instructions for the agent to ignore or neutralize embedded directives found within the retrieved case data.
Capability inventory: The skill possesses access to the Bash native tool and the SN-Add-Work-Notes MCP tool, which could be misused if the agent executes instructions embedded in case records.
Sanitization: The instructions do not define any automated sanitization, filtering, or validation steps for the content fetched from the ServiceNow instance before it is analyzed by the agent.

hrsd-case-summarization