← 返回 Skills 市场
337
总下载
1
收藏
1
当前安装
2
版本数
在 OpenClaw 中安装
/install clawvisual
功能描述
URL or long-form text to social carousel generator via local CLI + MCP endpoint.
使用说明 (SKILL.md)
clawvisual
Use clawvisual as a callable skill for agent workflows. It can auto-start a local web/MCP service and expose stable CLI commands for generation and job polling.
Quick start
npm install -g clawvisual
clawvisual set CLAWVISUAL_LLM_API_KEY "your_openrouter_key"
clawvisual initialize
clawvisual convert --input "https://example.com/article" --slides auto
clawvisual status --job \x3Cjob_id>
For local repo usage, you can also run:
npm run skill:clawvisual -- initialize
What it does
- Convert URL or direct long-form text into social carousel output.
- Provide job-based async workflow (
convert->status --job). - Support revision operations (
revise,regenerate-cover). - Expose MCP JSON-RPC tools for OpenClaw and other agent runtimes.
Commands
clawvisual initialize: probe/start local service and print Web URL.clawvisual status: check service identity (must beclawvisual).clawvisual tools: list MCP tools.clawvisual convert --input \x3Ctext_or_url> [--slides auto|1-8] [--ratio 4:5|1:1|9:16|16:9] [--lang \x3Ccode>]clawvisual status --job \x3Cjob_id>: query job state and result.clawvisual revise --job \x3Cjob_id> --instruction \x3Ctext> [--intent rewrite_copy_style|regenerate_cover|regenerate_slides]clawvisual regenerate-cover (--job \x3Cjob_id> [--instruction \x3Ctext>] | --prompt \x3Ctext>) [--ratio 4:5|1:1|9:16|16:9]clawvisual call --name \x3Ctool_name> --args \x3Cjson>: raw tool invocation.
Config
Optional local config file:
~/.clawvisual/config.json
Manage config with CLI:
clawvisual set CLAWVISUAL_LLM_API_KEY "your_key"
clawvisual get CLAWVISUAL_LLM_API_KEY
clawvisual config
clawvisual unset CLAWVISUAL_LLM_API_KEY
Supported keys:
CLAWVISUAL_LLM_API_KEY(alias:LLM_API_KEY)CLAWVISUAL_LLM_API_URL(alias:LLM_API_URL)CLAWVISUAL_LLM_MODEL(alias:LLM_MODEL)CLAWVISUAL_MCP_URL(alias:MCP_URL)CLAWVISUAL_API_KEY
When auto-starting local service, CLAWVISUAL_LLM_* values are mapped to runtime LLM_* envs.
Workflow pattern
initializeconvert- Poll
status --job \x3Cjob_id>until completion - Optional
revise/regenerate-cover - Poll revised job with
status --job
Output contract
All commands return JSON to stdout for deterministic parsing by upstream agents.
安全使用建议
Before installing or invoking this skill: 1) Confirm the npm package's provenance (check the package page, author, repository and recent publish activity) and inspect its source code if possible; 2) Prefer installing in an isolated environment (container or VM) until you trust it; 3) Use a dedicated, limited-scope API key for any LLM provider (do not reuse high-privilege or personal keys); 4) Be aware the tool will create ~/.clawvisual/config.json and can start a local web/MCP service — review/limit network access and file permissions; 5) If you need stronger control, avoid global install (-g) and do not provide secrets until you have reviewed the package; 6) If you want to proceed but remain cautious, run the CLI interactively and inspect its network calls (or run behind a local proxy) rather than giving the agent autonomous access to the skill.
功能分析
Type: OpenClaw Skill
Name: clawvisual
Version: 1.0.2
The clawvisual skill is a standard CLI wrapper for a social media carousel generator. It provides commands for managing LLM API keys, initializing a local service, and processing content via an external npm package ('clawvisual'). The documentation in SKILL.md describes a legitimate workflow for content generation and configuration management without any indicators of data exfiltration, malicious execution, or prompt injection.
能力评估
Purpose & Capability
The declared requirement (a 'clawvisual' binary) matches the skill's purpose (local CLI + MCP service for generating carousels). However, the registry metadata lists no required environment variables or config paths while the SKILL.md documents multiple environment keys (CLAWVISUAL_LLM_API_KEY, CLAWVISUAL_API_KEY, CLAWVISUAL_MCP_URL, etc.) and a config file (~/.clawvisual/config.json). That discrepancy is unexplained.
Instruction Scope
SKILL.md instructs the agent/user to install the npm package, set LLM and API keys, initialize a local web/MCP service, and then convert/poll/revise jobs. Those steps are consistent with a visual generation tool, but they also cause the agent to: (a) store/consume secret API keys, (b) start a local service that exposes MCP JSON-RPC tools (which other agents or processes could call), and (c) map CLAWVISUAL_LLM_* variables to runtime LLM_* envs (sending user content to external LLM endpoints). The SKILL.md also references the user's home config path even though the registry said none are required.
Install Mechanism
The registry package is instruction-only (no install spec), but the SKILL.md includes an npm install recommendation (npm install -g clawvisual) and a metadata install hint. That is a low-to-moderate risk pattern because installation would pull a package from npm; the registry did not provide a signed/verified install manifest. Verify package provenance before running npm -g.
Credentials
Although the registry lists no required env vars, the SKILL.md documents multiple secret-bearing env keys (LLM API keys, an API_KEY, MCP_URL). Asking for LLM/API keys is expected for a tool that calls external LLMs, but the lack of declaration in metadata is an inconsistency. Providing these keys would allow the tool to transmit content to third-party endpoints; that is proportionate only if you trust the package and the destination.
Persistence & Privilege
The skill does not request 'always: true' and is user-invocable (normal). It does create/use a per-user config file (~/.clawvisual/config.json) and can auto-start a local web/MCP service — a reasonable persistence model for a local CLI tool but one that increases the attack surface because local RPC endpoints can be invoked by other processes.
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 clawvisual - 安装完成后，直接呼叫该 Skill 的名称或使用
/clawvisual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
v1.0.2
- Service identity requirement updated: `clawvisual status` now expects identity to be `clawvisual` (was `clawvisual-mcp`)
- Documentation updated to reflect this change in command example and usage
v1.0.1
clawvisual 1.0.1
- Adds detailed documentation in SKILL.md for usage, setup, config, and workflow.
- Describes supported commands for converting URLs or text into social carousels via CLI and MCP endpoint.
- Lists optional configuration options and example CLI usage.
- Clarifies job-based async workflow and command JSON output contract.
- No changes to code or implementation details specified.
元数据
常见问题
Clawvisual 是什么？
URL or long-form text to social carousel generator via local CLI + MCP endpoint. 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 337 次。
如何安装 Clawvisual？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install clawvisual」即可一键安装，无需额外配置。
Clawvisual 是免费的吗？
是的，Clawvisual 完全免费，采用 MIT-0 许可证，可自由下载、安装和使用。
Clawvisual 支持哪些平台？
Clawvisual 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 Clawvisual？
由 suinia（@suinia）开发并维护，当前版本 v1.0.2。
推荐 Skills