The Agent Skills Directory

Indirect Prompt Injection (LOW): The skill identifies issues based on the output of the /check-stripe command and subsequently performs file system modifications and shell command execution.
Ingestion points: Output from the /check-stripe tool (referenced in the Process section).
Boundary markers: Absent; the skill does not specify delimiters for the input data.
Capability inventory: File system writes (updating TypeScript/Next.js files), git command execution, stripe CLI execution, and curl execution.
Sanitization: Absent; the skill lacks explicit validation or escaping of the ingested issue descriptions before acting upon them.
Data Exposure & Exfiltration (SAFE): The skill specifically targets and remediates hardcoded secrets (sk_test_...) by replacing them with environment variables (process.env.STRIPE_SECRET_KEY), which is a security best practice. All secrets shown in the documentation are non-functional placeholders.
Command Execution (SAFE): Uses standard development tools including git, curl (targeting localhost), and the official stripe CLI for verification purposes.

fix-stripe