RePrompter
PassAudited by ClawScan on May 10, 2026.
Overview
The skill appears purpose-aligned and not malicious, but its optional team mode can launch tmux-based agent teams and use local project context, so users should invoke it deliberately.
This looks safe to install for prompt rewriting. Treat Repromptception/team mode as active orchestration: use it only on projects where you are comfortable with agent execution, review the generated plans first, and disable project context when working with sensitive code or private details.
Findings (4)
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
If invoked on a real project, the skill may coordinate several agents that produce or execute work rather than only returning a cleaned-up prompt.
This shows the optional advanced mode can orchestrate multiple agents and retries through tmux. That is higher-impact than simple prompt rewriting, but it is disclosed as a named mode.
Repromptception ... Plan team → reprompt each agent → tmux Agent Teams → evaluate → retry
Use Repromptception only when you intend agent-team execution, review the generated team brief and per-agent prompts, and run it in a safe branch or workspace.
The advanced mode may fail or behave unexpectedly unless the local environment is intentionally prepared for Claude Code agent teams.
The advanced mode depends on tmux and an environment flag, while the registry requirements list no required binaries or environment variables. This is a setup/metadata gap, not evidence of hidden installation.
Repromptception mode requires Claude Code with tmux and CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS=1.
Before using team mode, confirm tmux and the experimental agent-teams setting are intentionally enabled; otherwise use the single prompt-improvement mode.
Project details such as framework, TypeScript, or database setup may appear in generated prompts or team briefs.
The skill can incorporate local project metadata into generated prompts. The artifacts also document transparency and a current-directory boundary, which keeps this proportionate.
Auto-detects tech stack and includes in context... Source transparency: "Auto-detected from: [pwd]"... No parent directory scanning
Review generated context before sharing or executing prompts, and use the documented no-context option for sensitive projects.
A team run may continue in tmux panes while agents work, so users should monitor or stop sessions if behavior is unexpected.
tmux-based parallel agents are a form of persistent session during team execution. The behavior is documented for the intended mode, with no evidence of hidden background activity.
Phase 3: tmux session created, agents execute in parallel
Keep tmux sessions visible during team runs and clean up or terminate any session you no longer want running.