The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill instructions guide the agent to ingest and process data from untrusted external platforms, creating a vulnerability surface for indirect prompt injection.
Ingestion points: The instruction files examples/3p-updates.md, examples/company-newsletter.md, and examples/faq-answers.md explicitly direct the agent to use Slack messages, Google Drive documents, emails, and calendar entries as data sources.
Boundary markers: The instructions do not define clear delimiters or specific instructions for the agent to ignore commands that may be embedded within the external documents it reads.
Capability inventory: The skill focuses on reading, summarizing, and formatting internal data. While the skill does not define its own dangerous tools, it operates in an environment where the agent may have access to other tools that could be abused if an injection succeeds.
Sanitization: There are no requirements for the agent to validate, filter, or sanitize the content retrieved from external tools before incorporating it into its final output.

internal-comms