qp-run
Pass
Audited by Gen Agent Trust Hub on Sep 6, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill invokes a local bash script (
plugins/fh-qp/scripts/qp_tools.sh) to perform specialized tasks including generating screen identifiers, checking MTM state, and masking sensitive content. - [INDIRECT_PROMPT_INJECTION]: The skill processes external data sources such as test plans (
tcs.tsv) and live application content (accessibility trees, screenshots), which represents a potential attack surface for indirect prompt injection. - Ingestion points: Reads
qp/plan/tcs.tsv,qp/plan/inventory.md, and captures real-time application state. - Boundary markers: The instructions define rigid output formats (TSV) and validation steps (
mtm-check), which act as structural constraints. - Capability inventory: Uses
BashandWriteto generate artifacts and execute local scripts. - Sanitization: Includes a dedicated masking step (
qp_tools.sh mask) to redact sensitive patterns (emails, tokens) from output evidence.
Audit Metadata