← 返回 Skills 市场
Claw Canvas
作者
austindixson
· GitHub ↗
· v1.0.0
514
总下载
0
收藏
1
当前安装
1
版本数
在 OpenClaw 中安装
/install claw-canvas
功能描述
A virtual canvas for OpenClaw to output content and visualize its thinking during development.
使用说明 (SKILL.md)
Claw Canvas Skill
Description
This skill wraps OpenClaw's native canvas tool to provide a dedicated, interactive surface for the agent to visualize its internal processes, display intermediate results, and output rich content directly on a virtual canvas. It enables a more transparent and intuitive development workflow by making the agent's thinking and work-in-progress visible.
Core Functionalities
- Render Markdown/HTML: Display formatted text, code, tables, and images.
- Visualize Data: Present charts, graphs, or structured data.
- Show Progress: Update the canvas with real-time progress of tasks.
- Interactive Thinking: Optionally display thought processes or decision trees.
- Snapshot: Capture the current state of the canvas.
Usage
This is primarily an internal skill for the agent to use to illustrate its workflow. It will expose a CLI interface for displaying content.
# Example: Display markdown content
python3 scripts/canvas_cli.py display_markdown --content "# Agent Thinking\
\
Here's my current thought process..."
# Example: Display an image
python3 scripts/canvas_cli.py display_image --url "https://example.com/image.png"
Purpose
To enhance transparency, improve user understanding of complex agent processes, and provide a dynamic, real-time output area for development tasks. This will be invaluable for tasks like:
- Visualizing website structure during the Mac App conversion.
- Displaying drafted blog posts or tweets with formatting.
- Showing data analysis results.
- Illustrating program flow or architectural decisions.
安全使用建议
This appears to be a small, legitimate helper that forwards content to the agent's canvas API. Before installing: (1) verify what openclaw.default_api.canvas does in your runtime (does it store, forward, or expose canvas contents externally?), (2) note that the SKILL.md advertises image/progress features that the included script does not implement — if you need those, request or inspect an updated implementation, (3) test the CLI in a safe environment to confirm behavior and output destinations, and (4) if you are concerned about sensitive content being shown, avoid sending secrets to the canvas and confirm any viewers/servers that can access canvas contents.
功能分析
Type: OpenClaw Skill
Name: claw-canvas
Version: 1.0.0
The `scripts/canvas_cli.py` skill takes arbitrary string input via the `--content` argument and passes it directly to the `default_api.canvas` tool for rendering. As the `SKILL.md` indicates the canvas can 'Render Markdown/HTML', this direct passing of unsanitized user-controlled content creates a potential vulnerability (e.g., Cross-Site Scripting if HTML is rendered, or other forms of injection) if the underlying `canvas` tool does not robustly sanitize or sandbox the input. This lack of input sanitization is a risky capability, classifying the skill as suspicious rather than benign, but not malicious as there's no evidence of intentional harmful behavior by the skill itself.
能力评估
Purpose & Capability
The name/description match the code: the CLI sends markdown to openclaw.default_api.canvas to present content. However SKILL.md advertises additional features (display_image, progress, snapshot, interactive thinking) that are not implemented in scripts/canvas_cli.py (only display_markdown exists). This is an inconsistency but not evidence of malicious behavior.
Instruction Scope
Runtime instructions are narrowly scoped to invoking the CLI to display content. The SKILL.md examples include an image-display command that the included CLI does not implement; otherwise the instructions do not request files, credentials, or unrelated system data. The CLI calls default_api.canvas — review what that API does in your runtime because the script delegates output handling to it.
Install Mechanism
No install spec and only a small Python script are included. Nothing is downloaded or written during install by this package itself.
Credentials
No environment variables, credentials, or config paths are required. The only dependency is the presence of an openclaw.default_api object at runtime (expected for an OpenClaw skill).
Persistence & Privilege
The skill does not request always:true, does not modify other skills or global configuration, and has no install-time persistence instructions.
如何使用
- 确保已安装 OpenClaw（本地或 Docker 部署）
- 在对话框中输入安装命令：
/install claw-canvas - 安装完成后，直接呼叫该 Skill 的名称或使用
/claw-canvas触发 - 根据 Skill 的参数说明提供必要输入，即可获得结构化输出
版本历史
v1.0.0
Initial release of Claw Canvas, providing a virtual, interactive surface for agent visualization:
- Offers rendering of Markdown/HTML, images, charts, and tables.
- Displays agent's thinking and progress in real-time.
- Includes a CLI for displaying various content types on the canvas.
- Supports capturing canvas snapshots.
- Aims to enhance workflow transparency and output visualization during development.
元数据
常见问题
Claw Canvas 是什么？
A virtual canvas for OpenClaw to output content and visualize its thinking during development. 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 514 次。
如何安装 Claw Canvas？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install claw-canvas」即可一键安装，无需额外配置。
Claw Canvas 是免费的吗？
是的，Claw Canvas 完全免费（开源免费），可自由下载、安装和使用。
Claw Canvas 支持哪些平台？
Claw Canvas 跨平台运行，可在任意部署了 OpenClaw / Claude Code 的环境中使用（cross-platform）。
谁开发了 Claw Canvas？
由 austindixson（@austindixson）开发并维护，当前版本 v1.0.0。
推荐 Skills