自定义多agent

AdvisoryAudited by Static analysis on Apr 30, 2026.

Overview

No suspicious patterns detected.

Findings (0)

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

If used, the skill can add or remove reusable agent template files in the OpenClaw workspace.

Why it was flagged

The skill describes user-directed creation and deletion of local template files. This is scoped and purpose-aligned, but it is still a mutation of persistent workspace content.

Skill content
Add your own template, saved to user-templates/ ... Remove user template (cannot remove system templates)
Recommendation

Use add/remove only when intended, and confirm the template ID or file path before deleting or overwriting a template.

What this means

A bad or untrusted template placed in the user template folder could steer future generated agents away from the behavior you expect.

Why it was flagged

User templates are persistent, automatically loaded, and can override built-in templates, so their prompt content can shape future generated agents.

Skill content
User Templates (Highest Priority) ... Override system templates with same ID ... Auto-detected
Recommendation

Keep only trusted templates in the template directory and review each prompt_template before reusing or sharing it.

What this means

Using the generated spawn snippet may start a separate agent session that receives the template prompt and any included task context.

Why it was flagged

The skill can generate instructions for spawning a separate agent session or thread using the selected prompt template. This is aligned with its stated purpose, but users should review what context is passed into the spawned agent.

Skill content
sessions_spawn( ... task: """{prompt_template}""", mode: "session", thread: true )
Recommendation

Review the generated sessions_spawn command before using it and avoid including secrets or unnecessary private information in spawned-agent prompts.