The Agent Skills Directory

[SAFE]: The skill provides instructions for business logic and product planning, specifically for scoping an MVP using MoSCoW prioritization and person-day budgeting.- [INDIRECT_PROMPT_INJECTION]: The skill prompts the agent to read repository files such as the README, package manifests, and data schemas to establish project context. This creates an ingestion surface for untrusted data; however, the skill lacks dangerous capabilities such as network access or shell execution, neutralizing the risk of a successful injection attack.- [DATA_EXPOSURE_AND_EXFILTRATION]: No patterns of data exfiltration or sensitive file access (e.g., to .env or .ssh) were found. The skill promotes secure practices by instructing the user to delegate authentication rather than implementing custom sign-in logic.- [REMOTE_CODE_EXECUTION]: There are no commands or instructions that download and execute code from remote sources. All instructions pertain to documentation and analysis within the local repository.

mvp