The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection attacks because it ingests untrusted user data to perform code generation and file system modifications using the agent's internal tools.
Ingestion points: The skill triggers and processes user input when phrases such as "graphql schema generator", "graphql generator", or "graphql" are detected.
Boundary markers: The skill instructions do not specify any boundary markers (e.g., XML tags or delimiters) to separate user-provided data from system instructions, which could allow the agent to misinterpret data as commands.
Capability inventory: The skill utilizes the Write, Edit, and Bash(curl:*) tools, providing a capability to modify files or perform network operations based on the generated outputs.
Sanitization: No input validation, filtering, or sanitization steps are defined to protect against malicious payloads within user requests.
[NO_CODE]: The skill package is composed entirely of the SKILL.md markdown file. It does not contain any executable scripts, binaries, or complex configuration files, which significantly reduces the risk of traditional remote code execution or malware inclusion.

graphql-schema-generator