The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted content from various repository files, which could contain malicious instructions designed to influence the agent's behavior during the triage process.
Ingestion points: The skill reads files from production/qa/bugs/, production/sprints/, and production/tasks/*/contract.md to extract bug details, task objectives, and acceptance criteria.
Boundary markers: The instructions do not define boundary markers (e.g., XML tags or clear delimiters) to separate the untrusted bug report content from the skill's system instructions, nor do they instruct the agent to ignore prompt-like content within those files.
Capability inventory: The skill utilizes Read, Glob, Grep, Write, and Edit tools. It has the capability to write a triage report to production/qa/bug-triage-[date].md and suggests the execution of further commands like /task-cycle or /task-bug based on ingested data.
Sanitization: The skill interpolates raw observations from bug reports directly into report templates and suggested command strings without explicit sanitization or escaping mechanisms.

bug-triage