The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill's operating rules direct the agent to suggest the command npx skills add wondelai/skills/<slug> --global to the user if specific sub-skills are missing. This involves the download and execution of code from an external, unverified source.
[EXTERNAL_DOWNLOADS]: The workflow relies on fetching additional components from the wondelai/skills/ repository, which is not a recognized trusted vendor and falls outside the author's identified infrastructure.
[COMMAND_EXECUTION]: The agent is instructed to propose shell commands to the user, including global package installations which may require elevated system privileges.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection through its session resumption mechanism. 1. Ingestion points: The agent reads docs/CREATE-APP-PLAN.md and all files listed in the Journey Map (e.g., PRODUCT.md, ARCHITECTURE.md) when starting a session. 2. Boundary markers: Absent; there are no instructions for the agent to distinguish between data and instructions within these files. 3. Capability inventory: The skill has extensive file-writing capabilities and can suggest shell command execution. 4. Sanitization: Absent; the skill does not perform any validation or filtering on the content of the read artifacts.

create-app