The Agent Skills Directory

[SAFE]: The skill serves as a security-focused guide for full-stack development. Its core instructions explicitly mandate security measures such as parameterized queries to prevent SQL injection, output encoding to mitigate XSS, and server-side validation.
[SAFE]: All external references are directed toward standard documentation domains (example.com), well-known cloud providers, or official package registries for established frameworks (FastAPI, NestJS, React).
[COMMAND_EXECUTION]: The skill includes instructional examples of standard development commands, such as npm ci within a multi-stage Dockerfile and GitHub Actions workflow. These are routine operations for building and testing software in a CI/CD context and do not include any remote execution patterns from untrusted sources.
[INDIRECT_PROMPT_INJECTION]: The skill maintains a surface for processing user-provided feature requirements to generate implementation plans.
Ingestion points: Technical requirements and acceptance criteria gathered in the first step of the workflow (SKILL.md).
Boundary markers: The process includes a mandatory 'Security checkpoint' and a 'Technical design' phase (specs/{feature}_design.md) which act as analytical gates before code generation.
Capability inventory: The skill generates structural code for web applications (models, controllers, components) but does not execute this code itself.
Sanitization: The skill's primary function is to instruct the user to implement sanitization using tools like Zod and Pydantic as part of the development lifecycle.

fullstack-guardian