The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill has an indirect prompt injection surface because it reads and displays content from files within the project workspace that could be modified by external actors or other processes.
Ingestion points: The agent is instructed to read component and screen definitions from ./fluxwing/components/, ./fluxwing/screens/, and ./fluxwing/library/ as documented in SKILL.md.
Boundary markers: There are no explicit instructions or delimiters provided to the agent to treat content from these files as untrusted or to ignore embedded instructions.
Capability inventory: The skill is configured with Read, Glob, and Grep capabilities, and its instructions imply the ability to perform file copy/write operations to move templates into the project library.
Sanitization: The skill lacks any content validation or sanitization requirements for the files it processes from the workspace.
[DATA_EXFILTRATION]: The skill uses path traversal ({SKILL_ROOT}/../) to access templates in sibling directories. While these paths (uxscii-component-creator and uxscii-screen-scaffolder) appear to be related functional components of the same toolset, traversing outside the skill root is a pattern that can lead to unintended data exposure if the environment is not strictly scoped.

Fluxwing Library Browser