The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes data from QuickBooks Online entities such as customers, invoices, and reports, which could potentially contain malicious instructions intended to influence the agent's behavior.
Ingestion points: Data is retrieved via qbo list, qbo get, qbo query, and qbo report commands (SKILL.md, references/COMMANDS.md).
Boundary markers: The instructions recommend using the --json flag and parsing with jq, which provides a degree of separation between data and execution logic.
Capability inventory: The skill allows for network operations via the qbo CLI and data modification (create, update, delete) within the QuickBooks environment.
Sanitization: The documentation does not specify sanitization methods for the retrieved data before it is presented to the agent.
[COMMAND_EXECUTION]: The skill relies on the qbo CLI tool to perform its operations. The agent is granted permission to run any qbo command via the Bash(qbo *) tool.
[EXTERNAL_DOWNLOADS]: The skill provides installation instructions for the qbo CLI using standard package managers (Homebrew, Scoop, Go) and direct binary downloads from the author's GitHub repository. These sources are considered safe as they are associated with the skill's author.

qbo