The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest user-provided requirements to generate system design documents, which theoretically creates an attack surface for indirect prompt injection if the user input contains malicious instructions.
Ingestion points: Phase 1 (Requirements Understanding) involves collecting user goals, actors, and constraints from SKILL.md.
Boundary markers: There are no explicit delimiters or specific instructions for the agent to isolate and ignore instructions embedded within the user's requirements.
Capability inventory: The skill is limited to writing markdown and Mermaid files to the docs/design-catalog/ directory using standard agent output or tools like TodoWrite.
Sanitization: The instructions do not specify a sanitization process for external content before it is included in design artifacts.
[NO_CODE]: The skill is entirely instructional and does not include, reference, or execute any local scripts, binaries, or external dependencies. It relies on the agent's internal capabilities to generate text and diagrams.
[SAFE]: The skill explicitly forbids the inclusion of implementation details such as database schemas, deployment configurations, or code examples, which reduces the risk of sensitive system information being generated or exposed.

system-design