The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the pscale command-line interface to interact with PlanetScale's billing API. This is the primary mechanism for all described tasks, including viewing payment methods, managing Stripe checkout flows, and listing invoices.
[DATA_EXFILTRATION]: While the skill accesses sensitive billing, payment, and invoice data, it includes explicit instructions to handle this data carefully. The agent is instructed not to paste raw invoice or payment-method output into public logs or chat without user approval, and to redact or aggregate details before sharing them outside the billing context.
[INDIRECT_PROMPT_INJECTION]: The skill processes external data returned from the pscale CLI (such as invoice line items and payment setup status).
Ingestion points: Untrusted data enters the agent context via the output of pscale billing invoice list, pscale billing invoice show, and pscale billing invoice line-items (specified in SKILL.md).
Boundary markers: The skill relies on natural language instructions for the agent to aggregate and redact details rather than formal delimiters.
Capability inventory: The skill uses subprocess calls to execute pscale commands (specified in SKILL.md and references/commands.md).
Sanitization: The instructions explicitly direct the agent to "redact or aggregate details before sharing outside the billing context" and "summarize costs at the level the user asked for."

pscale-billing