rerun-blueprint
Pass
Audited by Gen Agent Trust Hub on Sep 1, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill contains standard documentation for the Rerun visualization library. The provided Python snippets demonstrate how to interact with the SDK to visualize recordings, which is the intended purpose of the skill.
- [COMMAND_EXECUTION]: Code examples demonstrate spawning a local viewer process and managing gRPC connections for data streaming. These are standard operations for the Rerun SDK and do not involve unauthorized command execution.
- [INDIRECT_PROMPT_INJECTION]: The skill documents the ingestion of data from RRD files (segment.rrd) using RrdReader and log_file_from_path in SKILL.md. While these represent ingestion points for external data, the skill's logic is limited to visualization mapping and does not expose a surface for executing instructions embedded within the data. The tool capability inventory includes Bash, Read, and WebFetch, but no boundary markers or explicit sanitization steps are defined in the instructions, which is acceptable for this visualization use case.
Audit Metadata