The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill reads external markdown files and interpolates their content into generated files and summaries, creating a potential vector for instructions embedded in data to influence agent behavior.
Ingestion points: The skill reads extensively from design/gdd/*.md, docs/architecture/architecture.md, docs/architecture/tr-registry.yaml, and various Architecture Decision Records (ADRs).
Boundary markers: There are no explicit delimiters or system instructions directing the agent to ignore potentially malicious commands embedded within the GDD or ADR text during processing.
Capability inventory: The skill utilizes Read, Write, Grep, Glob, and Task tools. This allows the agent to modify the project structure and trigger sub-tasks (like the producer gate) based on input data.
Sanitization: The skill lacks explicit sanitization or validation logic for the text extracted from design documents before it is used to populate EPIC.md files or presented to the user.

create-epics