The Agent Skills Directory

[SAFE]: The skill instructions and associated files do not contain any malicious patterns, prompt injections, or attempts to bypass safety guidelines.
[COMMAND_EXECUTION]: The skill references a local Python script scripts/check-output.py used to validate generated deliverables. The script is benign, performing simple keyword checks on local text files without network access or unsafe execution patterns.
[DATA_EXPOSURE]: No hardcoded credentials, sensitive file paths, or data exfiltration patterns were identified. All recommended tools (e.g., Stripe, Clerk, Sentry) are industry-standard services mentioned in the context of SaaS development.
[INDIRECT_PROMPT_INJECTION]: While the skill ingests user descriptions of SaaS ideas, it does not have high-privilege capabilities or network access that could be exploited via injected data. The risk is negligible.

building-saas