The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and act upon external documentation and existing repository material, which introduces a surface for indirect instructions to influence the agent.
Ingestion points: The workflow in SKILL.md explicitly instructs the agent to inspect available files, repository evidence, and external documentation via sandbase_discover and sandbase_inspect.
Boundary markers: The instructions lack explicit boundary markers or commands to ignore instructions embedded within the external data sources.
Capability inventory: The skill utilizes the sandbase_run tool (documented in references/sandbase-api-map.md) which allows for the execution of discovered capabilities, representing a potential execution path for instructions found in processed data.
Sanitization: There is no evidence of sanitization, escaping, or filtering of the external content before it is processed by the agent.

chrome-devtools