The Agent Skills Directory

[PROMPT_INJECTION]: The skill is designed to ingest and process untrusted external blog content (Markdown, MDX, HTML), which creates a surface for indirect prompt injection. A source file containing malicious instructions could potentially influence the agent's behavior during the translation process.\n
Ingestion points: SKILL.md Phase 1 describes reading source files provided by the user.\n
Boundary markers: Absent; the instructions do not specify the use of delimiters (e.g., XML tags or triple quotes) or explicit warnings for the agent to ignore instructions embedded within the source content.\n
Capability inventory: The skill performs file read and write operations on the local filesystem (SKILL.md Phase 1 and Phase 5).\n
Sanitization: Absent; the skill includes checks for translation quality and artifacts but lacks mechanisms to detect or sanitize malicious prompt injection payloads within the input text.

blog-translate