dmux-workflows
Pass
Audited by Gen Agent Trust Hub on Jul 15, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill instructs the user to install the
dmuxutility globally via NPM (npm install -g dmux) and provides a reference to its GitHub repository (github.com/standardagents/dmux). These references are consistent with the skill's stated purpose of orchestration. - [COMMAND_EXECUTION]: The skill utilizes a local script
scripts/orchestrate-worktrees.jsto manage git worktrees and automate tmux session creation. It also supports alauncherCommandconfiguration that allows for the execution of arbitrary shell commands within managed agent panes. - [PROMPT_INJECTION]: The skill describes multi-agent feedback patterns (e.g., 'Test + Fix Loop') where one agent processes the output of another. This creates an indirect prompt injection surface.
- Ingestion points: Agents reading files like
handoff.md,status.md, or captured tmux pane output from other agents (SKILL.md). - Boundary markers: None explicitly defined in the workflow patterns.
- Capability inventory: Execution of shell commands via
launcherCommandand orchestration scripts (SKILL.md). - Sanitization: No sanitization or verification of inter-agent communication is described.
Audit Metadata