The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to interact with a user's codebase using tools like Read, Grep, and Glob. Because it also has high-privilege capabilities like Write, Edit, and Bash, there is a theoretical surface for indirect prompt injection if the agent processes source code containing malicious instructions (e.g., hidden in code comments or data schemas).
Ingestion points: Local source code files accessed via Read, Grep, and Glob tools.
Boundary markers: The instructions do not define specific delimiters or instructions to ignore embedded content within the files it analyzes.
Capability inventory: The skill allows file modification (Write, Edit) and shell execution (Bash).
Sanitization: No explicit sanitization or input validation for ingested code content is present in the skill configuration.

graphql-architect