The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user data when evaluating worker classification, which creates a potential surface for indirect prompt injection.
Ingestion points: The skill ingests user-provided workday descriptions, contract terms, and behavioral control facts in SKILL.md (Workflow Step 1) and references/common-law-test.md to perform the common-law classification test.
Boundary markers: There are no explicit boundary markers or "ignore embedded instructions" warnings defined for the interpolation of these user-provided work descriptions into the agent's reasoning process.
Capability inventory: The skill has capabilities for browser automation and filing (filing.md), as well as generating structured tax drafts that the user is instructed to copy into official filing software.
Sanitization: The skill lacks explicit instructions for sanitizing or escaping the user's narrative work descriptions before they are analyzed for tax classification markers.

form-8919