The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process data from external files, which creates a potential surface for indirect prompt injection if those files contain malicious instructions.
Ingestion points: As described in instructions.md, the agent is instructed to accept and process subcontractor data from CSV, Excel, and JSON formats provided by the user.
Boundary markers: The skill documentation and implementation do not define specific delimiters or instructions to the agent to treat ingested content strictly as data or to ignore any commands embedded within the input files.
Capability inventory: The technical implementation in SKILL.md utilizes the pandas library, which provides the ability to perform filesystem operations such as writing Excel files (to_excel).
Sanitization: The skill lacks explicit sanitization or filtering logic to inspect user-provided data for potentially malicious prompt content before processing it.

subcontractor-prequalification