The Agent Skills Directory

[SAFE]: The skill defines a structured and benign process for development task decomposition.- [DATA_EXFILTRATION]: The skill instructs the agent to read issues and explore codebase files. This access is necessary for the stated purpose of task analysis and does not involve any instructions to transmit data to external or unauthorized systems.- [COMMAND_EXECUTION]: The skill involves exploring the local filesystem and writing a task file. These are standard operations for a developer assistant and do not include any high-risk shell commands, privilege escalation, or unauthorized modifications.- [PROMPT_INJECTION]: The skill identifies a potential surface for indirect prompt injection (Category 8) as it processes issue descriptions from external sources such as files or URLs. 1. Ingestion points: Issue descriptions provided via URL or file referenced in SKILL.md. 2. Boundary markers: None explicitly defined to separate untrusted issue content. 3. Capability inventory: File system traversal, reading local files, and writing task files. 4. Sanitization: Not mentioned, however, the skill incorporates a mandatory 'Quiz the user' step for review and approval before any task list is finalized, which serves as a human-in-the-loop security control.

issues-to-tasks