The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted content from the user's filesystem and artifact registry, creating a potential surface for instructions hidden within the audited code to influence agent behavior.
Ingestion points: The skill reads target files via the [file-path] argument and resolves artifacts from .styleseed/bundles/ and .styleseed/manifests/.
Boundary markers: Absent. There are no explicit instructions or delimiters defined to prevent the agent from following instructions potentially embedded in the source code being audited.
Capability inventory: The skill is granted Write and Edit permissions, allowing the agent to modify files based on its audit findings.
Sanitization: Absent. The agent performs a static code audit without sanitizing or escaping the content it reads before processing it.
[EXTERNAL_DOWNLOADS]: The skill references an external local file ../ss-update/references/update-preflight.md for pre-workflow instructions. This is a standard dependency within the project structure.

ss-a11y