The Agent Skills Directory

[PROMPT_INJECTION]: The skill defines an instruction priority where its internal logic overrides default system prompts. It also contains 'Red Flags' (behavioral constraints) designed to prevent the model from bypassing the intended SDLC process. These instructions are functional directives for the agent's workflow and do not constitute an attempt to subvert safety guidelines or extract sensitive system information.
[INDIRECT_PROMPT_INJECTION]: The skill uses Jira MCP tools to ingest external data, creating a potential surface for indirect injection. • Ingestion points: Ticket descriptions and comments are read from Jira (SKILL.md). • Boundary markers: No explicit delimiters are used to separate retrieved Jira content from the agent's internal instructions. • Capability inventory: The skill routes to other components capable of executing shell commands and modifying files, such as 'dev-implement' and 'deploy-ship'. • Sanitization: The skill does not specify any sanitization or validation logic for the data retrieved via the Jira MCP interface.

enggenie