The Agent Skills Directory

[PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface by reading and summarizing content from arbitrary project files (documentation, code, etc.) without sanitization or boundary markers.
Ingestion points: Scans files across design/, src/, production/, prototypes/, docs/architecture/, and tests/ (identified in SKILL.md).
Boundary markers: The instructions do not define delimiters to separate analyzed content from the agent's own instructions when generating reports.
Capability inventory: The agent uses Read, Glob, Grep, and Bash to collect data and Write to create the final report.
Sanitization: The workflow lacks steps to escape or filter potentially malicious instructions contained within the project artifacts before they are summarized in the stage report template.

project-stage-detect