The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to read and process content from arbitrary specification documents (US, WI, FT, or external files) to extract acceptance criteria. This represents an attack surface where malicious instructions embedded within the criteria of a document could attempt to influence the agent's behavior during test case generation.
Ingestion points: SKILL.md Step 1 involves reading and extracting criteria from the README.md of artifacts or user-indicated external documents.
Boundary markers: The instructions do not specify the use of delimiters or 'ignore' instructions for the ingested content.
Capability inventory: The skill performs file system writes (test-cases/ directory) and makes external API calls to Azure DevOps via MCP tools.
Sanitization: No explicit sanitization or validation of the ingested text is performed before it is used to generate the test case titles and steps.
[DATA_EXFILTRATION]: If a repository is linked to Azure DevOps (configured in .agents/MEMORY.md), the skill synchronizes the full content of generated test cases to the external service using MCP tools. This includes the title, description, pre-conditions, and execution steps. This data transfer targets a well-known service (Microsoft Azure DevOps) and is a documented functional requirement of the skill.
[COMMAND_EXECUTION]: The skill utilizes external Model Context Protocol (MCP) tools (identified as mcp_azure-devops_*) to perform administrative tasks within Azure DevOps, such as searching for or creating Test Plans, Test Suites, and Test Case work items. These operations are restricted to the context of the configured Azure DevOps organization and project.

test-define