The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted JSON and JSONL data, which presents a surface for indirect prompt injection attacks if the data contains malicious instructions intended for the agent.
Ingestion points: The skill reads external data using json.load and streams JSONL files line-by-line (SKILL.md).
Boundary markers: The instructions do not explicitly mandate delimiters or warnings to ignore instructions embedded within the JSON data being processed.
Capability inventory: The skill utilizes file system access for reading and writing data, along with data manipulation libraries (SKILL.md).
Sanitization: No content-level sanitization or validation of the data values is described.
[EXTERNAL_DOWNLOADS]: The skill requires the installation of the pandas library via pip (SKILL.md). pandas is a standard, well-known library for data analysis and manipulation.
[COMMAND_EXECUTION]: The skill instructs the agent to run a shell command to install a dependency (SKILL.md).
Evidence: pip install pandas

json-restructure