The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process external backlog items which could potentially contain malicious instructions or hidden prompts.
Ingestion points: The skill reads backlog data from the workspace to classify and refine items as described in SKILL.md.
Boundary markers: The instructions do not define specific delimiters or instructions for the agent to distinguish between the skill's instructions and the content of the backlog items.
Capability inventory: The skill is restricted by sandbox_mode: read-only in the metadata and does not contain any executable scripts, tool calls, or network operations.
Sanitization: There is no evidence of content validation or sanitization applied to the ingested backlog data.

backlog-grooming