The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes a user-supplied pricing_description to determine how to configure billable metrics, products, and rate cards, which creates a potential surface for indirect prompt injection where malicious input could attempt to influence the generated API requests.
Ingestion points: Untrusted data enters the agent context through the pricing_description argument defined in SKILL.md.
Boundary markers: The skill mitigates this risk by requiring a "Confirmation gate" (as specified in references/billable-metrics.md), which forces the agent to present a proposed configuration table and wait for explicit user approval before executing any metric creation calls.
Capability inventory: The skill utilizes HTTP tools to perform POST and GET requests against the Metronome API endpoints at api.metronome.com for infrastructure setup and usage verification.
Sanitization: The skill relies on the agent's internal reasoning and the mandatory human review step to ensure the integrity of the proposed billing configuration.

metronome-setup-catalog