The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to process external project plans to generate tickets, which serves as an ingestion point for untrusted data. Although this creates a theoretical surface for indirect prompt injection, the skill includes strong safety mitigations. Ingestion points: Project plan descriptions and work items normalized during the core process in SKILL.md. Boundary markers: No explicit delimiters are used to isolate untrusted input, but the skill requires a specific five-section structured output. Capability inventory: Integration with tracker APIs (Jira, Linear, GitHub Issues) for ticket creation. Sanitization: The skill enforces a 'HARD-GATE' that requires explicit user confirmation, defaults to 'draft-only' mode, and mandates the validation of a single issue before bulk creation.

plan-tickets