The Agent Skills Directory

Indirect Prompt Injection Surface: The skill processes external customer, order, and carrier records (via classified.json and records.json) to consolidate context for service inquiries. This ingestion of external data represents a potential surface for indirect prompt injection if the source data contains instructions intended for the agent.
Ingestion points: The Python script scripts/context_assemble.py reads data from classified.json and records.json to generate the assembled.json output.
Boundary markers: The skill instructions do not specify explicit delimiters or boundary markers for the interpolated context data.
Capability inventory: The skill performs file read/write operations and executes a local Python script to perform deterministic logic.
Sanitization: While the script uses deterministic logic for path selection, it does not implement specific sanitization or filtering of the content within the input records.
Local Script Execution: The skill executes a local Python script (scripts/context_assemble.py) to process data. This is a standard pattern for performing deterministic business logic (such as checking loyalty tiers and order values) that would be difficult to perform reliably via natural language alone.

context-assemble