Docker Container Rerun

ReviewAudited by ClawScan on May 10, 2026.

Overview

The skill is transparent and scoped to updating a user-specified Docker container, but it can stop/remove/recreate containers and expose recent logs or inspect output.

Use this skill only when you intend to let the agent manage a specific Docker-run container. Provide a trusted original `docker run` command, review it exactly before approving `--apply`, and be aware that Docker inspect/log output can contain sensitive information.

Findings (3)

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 wrong container name or recreate command could cause downtime or remove the container's writable layer.

Why it was flagged

The skill explicitly performs Docker mutations that can stop and replace a running container. This is central to the stated purpose and is paired with approval guidance, but it is still high-impact.

Skill content
If Ids differ:\n   - run `docker stop <container_name>`\n   - run `docker rm <container_name>`\n   - run the exact `recreate_command`
Recommendation

Verify the target container and the exact recreate command before approving apply mode; ensure important data is stored in volumes or backed up.

What this means

If an unsafe or tampered recreate command is approved, it could run more than just the intended Docker command.

Why it was flagged

The bundled script executes the user-provided recreate command through Bash. This preserves the exact command, but any approved shell operators or substitutions in that command will also execute.

Skill content
proc = subprocess.run(recreate_command, shell=True, executable="/bin/bash", text=True)
Recommendation

Only approve a trusted full `docker run` command, and be cautious with pasted commands containing `;`, `&&`, `$()`, backticks, or unexpected redirects.

What this means

Sensitive operational details or secrets from the container may appear in the chat or tool output.

Why it was flagged

Container inspect output and recent logs can include environment variables, service metadata, application errors, or secrets, and may be brought into the agent conversation.

Skill content
Verify startup with:\n   ```bash\n   docker ps --filter name=<container_name>\n   docker inspect <container_name>\n   docker logs --tail 100 <container_name>\n   ```
Recommendation

Avoid sharing sensitive logs unnecessarily, and redact secrets before copying or storing results.