The Agent Skills Directory

[COMMAND_EXECUTION] (HIGH): The skill contains a shell escape command !cat .cargo/config.toml intended to inject local file content into the agent's context. This pattern represents arbitrary command execution on the host machine.
[DATA_EXFILTRATION] (MEDIUM): The skill targets .cargo/config.toml, which may contain sensitive data such as private registry credentials, auth tokens, or environment-specific paths. Accessing these files without explicit user consent constitutes a data exposure risk.
[INDIRECT_PROMPT_INJECTION] (MEDIUM): This skill creates an injection surface by reading external file content into the prompt.
Ingestion points: .cargo/config.toml (via cat in SKILL.md).
Boundary markers: Absent; the content is injected directly without delimiters.
Capability inventory: As an embedded development skill, the associated agent likely has high privileges including file system access and the ability to execute build commands (cargo).
Sanitization: Absent; the output of the shell command is piped directly into the context, allowing a malicious project file to influence agent behavior.

domain-embedded