The Agent Skills Directory

[SAFE]: The skill follows standard best practices for calculating business mileage deductions. It does not attempt to execute arbitrary code or exfiltrate sensitive information.
[INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied mileage data and transaction records.
Ingestion points: User input for miles driven; tool output from transaction_search (SKILL.md).
Boundary markers: None explicitly defined for the interpolation of data.
Capability inventory: The skill uses transaction_search and export_transactions. It lacks the ability to execute shell commands, write arbitrary files, or perform non-whitelisted network operations.
Sanitization: The skill instructs the agent to perform simple arithmetic (multiplication) and mapping to tax forms, which inherently limits the impact of malicious data injection.

mileage-tracker