Preqstation

ReviewAudited by ClawScan on May 10, 2026.

Overview

This skill is coherent for delegating coding work, but it explicitly allows sandbox/approval-bypass style CLI execution and background coding agents, so users should review it before use.

Install only if you want OpenClaw to delegate coding work to local Claude, Codex, or Gemini CLIs. Before use, verify the mapped project path, prefer sandboxed execution when possible, monitor background sessions, and review all worktree diffs before merging.

Findings (5)

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.

What this means

A delegated coding agent could run commands or change files with fewer safeguards than expected, even though the skill tries to keep work inside a git worktree.

Why it was flagged

The skill instructs the agent to use flags that may bypass normal sandboxing or approval safeguards for coding-agent execution.

Skill content
Use `dangerously-*` / sandbox-disable flags only for actual coding execution with local trusted CLIs.
Recommendation

Use only with trusted repositories and trusted CLI installations. Prefer sandboxed modes where possible, and require explicit user confirmation before any dangerous or sandbox-disabling flag is used.

What this means

Installing and invoking this skill can cause Claude Code, Codex CLI, or Gemini CLI to execute coding tasks locally.

Why it was flagged

The skill is designed to launch local coding CLI processes, which is expected for its purpose but still means it performs local command execution.

Skill content
create a per-task git worktree and use it as execution `<cwd>`; launch engine commands with `pty:true` and explicit `workdir:<cwd>`
Recommendation

Review the chosen worktree, inspect diffs before merging, and avoid using this on sensitive repositories unless you trust the selected CLI and its configuration.

What this means

A coding task may continue running in the background after the initial response, consuming resources or continuing to modify the worktree.

Why it was flagged

The skill intentionally starts long-running background coding sessions by default, though it also documents polling, logging, and kill controls.

Skill content
Launch coding agents with `background:true` by default. Use foreground only when user explicitly asks for blocking/synchronous execution.
Recommendation

Monitor background sessions with the documented process controls and explicitly request foreground execution for sensitive tasks.

What this means

Wrong, stale, or overly broad path mappings could send future coding tasks to the wrong repository, and absolute paths may reveal local workspace structure if shared.

Why it was flagged

The skill persists workspace path mappings that can affect where future coding agents are launched.

Skill content
This skill reads and updates `MEMORY.md` project mappings with absolute paths.
Recommendation

Keep MEMORY.md mappings minimal and review them before running tasks; only store intended project checkouts.

What this means

The package appears instruction-only, but version mismatch can make provenance and review tracking less clear.

Why it was flagged

The submitted registry metadata lists version 0.1.8, while the packaged _meta.json lists 0.1.5, indicating metadata drift.

Skill content
"version": "0.1.5"
Recommendation

Verify the intended repository and package version before relying on this skill in a sensitive workflow.