The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes user-provided EDI files, which serve as an untrusted data ingestion point. * Ingestion points: Files retrieved via /catalog or user attachments are processed in SKILL.md steps. * Boundary markers: None identified; the skill does not explicitly instruct the agent to ignore instructions within the EDI data. * Capability inventory: The skill uses the bash tool to run edi, csv, and jq commands, and writes to the /tmp directory. * Sanitization: No content validation or sanitization is performed on the EDI file before it is parsed and presented to the agent.

edi-remittance