lift
Pass
Audited by Gen Agent Trust Hub on Mar 2, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill downloads source code from the author's repository at
https://github.com/tkersey/skills-zigand installs packages via thetkersey/tap/liftHomebrew tap. These resources are vendor-controlled and used to set up the benchmarking environment. - [COMMAND_EXECUTION]: The skill performs command-line operations to build and execute its performance tools. This includes
git cloneto fetch source,zig buildto compile the tools locally, and direct execution of thebench_statsandperf_reportbinaries to gather metrics. - [PROMPT_INJECTION]: The skill identifies as an indirect prompt injection surface as it is designed to execute external workloads for benchmarking purposes.
- Ingestion points: The skill reads and runs user-provided reproduction scripts and workload commands as described in
SKILL.md. - Boundary markers: No explicit markers or delimiters are used to isolate the workload commands during execution.
- Capability inventory: The skill executes shell commands and compiled binaries with access to the local filesystem for report generation.
- Sanitization: As a performance tool, it executes workloads directly to capture realistic metrics and does not perform content validation on the input scripts.
Audit Metadata