The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill documentation includes instructions to run npm install, which fetches dependencies from the well-known npm registry.
[COMMAND_EXECUTION]: The skill relies on a shell script generate.sh and system binaries such as node, jq, and weasyprint to process JSON data and generate output files.
[PROMPT_INJECTION]: The skill contains an indirect prompt injection surface due to its handling of external JSON data.
Ingestion points: Data enters the agent context via stdin, local file paths, or as pre-saved configuration files in the $INVOICE_DIR/configs/ directory.
Boundary markers: No specific delimiters or instructions to ignore embedded commands are mentioned to protect against instruction injection within the JSON fields.
Capability inventory: The skill has the capability to execute shell scripts, perform filesystem operations, and run powerful command-line tools like jq and weasyprint.
Sanitization: The documentation does not specify any sanitization or validation logic used to prevent malicious input from affecting the behavior of the execution environment or subprocesses.

invoice-generator