skills/bmad-code-org/bmad-method-test-architecture-enterprise/bmad-testarch-trace/Gen Agent Trust Hub
bmad-testarch-trace
Warn
Audited by Gen Agent Trust Hub on Apr 26, 2026
Risk Level: MEDIUMCOMMAND_EXECUTIONPROMPT_INJECTIONDATA_EXFILTRATION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses local Python scripts (
resolve_customization.py) and various shell scripts (scripts/publish-pact.sh,scripts/check-pact-determinism.sh) to manage its workflow and integrate with testing tools. - [DYNAMIC_EXECUTION]: The workflow steps in
steps-v/step-01-validate.mdandsteps-e/step-02-apply-edit.mdinstruct the agent to execute arbitrary terminal commands resolved from theworkflow.on_completekey in configuration files. While intended for post-completion hooks, this provides a mechanism for arbitrary command execution if local configuration overrides are modified. - [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from the project root, such as Product Requirement Documents (PRDs), Tech Specs, and Acceptance Criteria (
step-01-load-context.md), to build its coverage oracle. These documents could contain instructions that attempt to influence the agent's gap analysis or gate decisions.
Audit Metadata