The Agent Skills Directory

[SAFE]: The skill is designed to perform UX audits on local files using a predefined framework. It adheres to the principle of least privilege by restricting its allowed-tools to Read, Grep, and Glob, preventing unauthorized network access or command execution.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from the local filesystem to perform its audit.
Ingestion points: The skill reads file contents from .styleseed/bundles/ and .styleseed/manifests/ based on user-provided arguments.
Boundary markers: Absent. The instructions do not define explicit delimiters to separate analyzed data from the agent's internal logic.
Capability inventory: The skill is limited to Read, Grep, and Glob. It lacks the ability to write files, execute shell commands, or make network requests.
Sanitization: Absent. The skill processes the raw text of the files provided.
Risk Assessment: While an attacker could place malicious instructions inside an audited file, the lack of dangerous capabilities (network/write/exec) restricts the impact to potentially biased audit results.

ss-audit