The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The scripts/agent.py script processes external data from JSON files to generate intelligence reports. If the input data contains malicious instructions, these could be rendered into the output report and potentially influence the agent's behavior during subsequent review or processing steps.
Ingestion points: Data is ingested from file paths provided to the --data argument in scripts/agent.py.
Boundary markers: The skill does not implement delimiters or 'ignore' instructions to isolate processed data from the agent's control flow.
Capability inventory: The script has the capability to write files to the local file system based on user-provided output directories.
Sanitization: Input data is interpolated directly into templates without validation or sanitization to remove potentially malicious commands or markdown injections.
[DYNAMIC_EXECUTION]: The skill utilizes the jinja2 templating engine to dynamically generate Markdown content at runtime. While the report templates are statically defined within the script, the process of assembling executable-like Markdown structures from external data represents a dynamic generation pattern.

generating-threat-intelligence-reports