The Agent Skills Directory

[SAFE]: The skill acts as a project-starter assistant. It follows a structured, multi-phase interview process to help users define a Minimum Viable Product (MVP) and core project components.
[DATA_EXPOSURE]: The skill interacts with the local file system by reading and writing to the learning/ directory (specifically learning/project.md). This behavior is transparent, documented in the skill's instructions, and limited to project-related metadata rather than sensitive system files or credentials.
[INDIRECT_PROMPT_INJECTION]: The skill processes user responses and reads existing local project files. This interaction is scoped to an interview context and does not involve risky operations like executing user-provided strings in a shell or making network requests, effectively mitigating the impact of potential prompt injection.

start-project