The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill reads and processes the content of various project files, which represents an attack surface where instructions could be embedded to influence the agent's behavior.
Ingestion points: The process involves reading .planning/ROADMAP.md and phase-specific files such as CONTEXT.md and RESEARCH.md.
Boundary markers: The skill does not define clear boundary markers or provide instructions to the agent to disregard any embedded commands within these files.
Capability inventory: The skill is capable of executing directory operations (list, rename, delete) via Bash, interacting with the gsd-sdk CLI tool, and modifying files via the Write tool.
Sanitization: There is no evidence of sanitization or strict schema validation for the content read from the markdown files before it is processed.

gsd-review-backlog