The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process data from external sources and generative models via the SandBase API, creating a surface for indirect prompt injection attacks.
Ingestion points: External data sources and model inference results are fetched using the sandbase_run tool as described in SKILL.md and references/sandbase-api-map.md.
Boundary markers: The instructions do not specify the use of delimiters (e.g., XML tags or triple quotes) or provide explicit warnings for the agent to ignore instructions that might be embedded within the external data.
Capability inventory: The skill utilizes sandbase_run for network operations to retrieve data and sandbase_inspect to evaluate schemas, which are standard for data integration but can be leveraged if malicious data is ingested.
Sanitization: There are no explicit instructions for sanitizing, escaping, or filtering external content to prevent instructions in the data from influencing the agent's behavior.

seaborn