The Agent Skills Directory

[SAFE]: The skill is exclusively designed for planning and documentation. It explicitly states it does not generate application code, execute tests, or create architectural decision records (ADRs) independently.
[SAFE]: Integration with Azure DevOps (ADO) is performed using local project metadata and standard Model Context Protocol (MCP) tools. The process involves creating work items first in ADO and then using their IDs for local file naming, which is a standard development workflow.
[PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection as it ingests content from repository files like user stories and existing task descriptions.
Ingestion points: Technical documentation files including README.md (User Stories), TK-*.md (Tasks), and WI-*.md (Work Items) are read to inform the planning process.
Boundary markers: The skill provides structural guidance by directing the agent to specific sections (e.g., 'Criterios de aceptación'), though it lacks explicit delimiters to ignore embedded instructions in those files.
Capability inventory: Capabilities are limited to creating markdown files in specific documentation directories and making tool calls to an Azure DevOps MCP. It lacks the ability to execute shell commands or arbitrary code.
Sanitization: The skill does not implement specific sanitization for the content read from markdown files, relying on the agent's internal safety filters.

work-plan