The Agent Skills Directory

[COMMAND_EXECUTION]: The skill requires the environment variable SAP_ALLOW_FREE_SQL=true to function. This allows the AI agent to execute arbitrary, unvalidated SQL queries (FreeSQL) against the SAP database, which is a high-privilege operation that can bypass traditional application-level security controls.
[DATA_EXFILTRATION]: The skill accesses SAP usage monitoring tables including SCMON_DATA and SUSG_I_DATA. These tables contain technical metadata about system execution and architecture. While they do not typically store business data, access to this execution telemetry provides a map of the system's internal logic and usage density.
[EXTERNAL_DOWNLOADS]: The skill references a repository on AWS's official GitHub account (aws-solutions-library-samples) as a recommended tool for processing SUSG XML exports. This is a reference to a well-known and reputable technology service provider.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection (SQL injection) because it dynamically generates SQL queries using user-provided package names, namespaces, and object lists without explicitly instructing the agent to sanitize or validate these inputs.
Ingestion points: User-provided inputs for 'Package', 'Namespace prefix', and 'Object list' (File: SKILL.md).
Boundary markers: The instructions do not define boundary markers or escaping logic for user strings used in SQL WHERE clauses.
Capability inventory: The skill utilizes SAPQuery for FreeSQL execution, alongside SAPRead, SAPSearch, and SAPNavigate tools (File: SKILL.md).
Sanitization: There are no instructions for the agent to sanitize or validate the user-supplied strings before interpolating them into executable SQL commands.

sap-unused-code