The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from the conversation history and external reports to generate summaries and visualizations.
Ingestion points: The agent is instructed to collect its own work, reports from subagents, and messages from other users in the chat history as described in SKILL.md.
Boundary markers: There are no instructions to use delimiters or specific warnings to ignore instructions that might be embedded within the ingested chat content.
Capability inventory: The skill uses html_preview and html_render to create visual summaries and refers to local skill definitions in sibling directories (e.g., ../<name>/SKILL.md).
Sanitization: The instructions do not include steps to sanitize or escape the content gathered from the chat history before it is used in the summary or rendered in visual components.

ketchup