← 返回 Skills 市场
Toggle
作者
Aleksandar Yordanov
· GitHub ↗
· v1.0.6
946
总下载
8
收藏
2
当前安装
7
版本数
在 OpenClaw 中安装
/install toggle
功能描述
Context layer for your agent. ToggleX captures the user's work sessions, projects, focus scores, and context switches across the web — giving the agent aware...
安全使用建议
This skill appears coherent for its stated purpose, but consider the following before installing:
- Endpoint trust: the script sends your TOGGLE_API_KEY to https://ai-x.toggle.pro. Only use a key if you trust that service. Verify the domain and integration page before adding a real API key.
- Least privilege: create a dedicated ToggleX integration key with limited scope/expiry if possible, rather than using a broad or long-lived user key.
- Memory persistence: when you run with --persist (or use the suggested ../../memory path), the skill will write markdown files containing your activity data. Confirm where those files will be stored and who/what can read them.
- Cron checks: the script reads ~/.openclaw/cron/jobs.json to check scheduling. It will create state.yaml inside the skill dir if missing. Review that file before trusting automated runs.
- Downstream automation caution: SKILL.md encourages proactive nudges and automations. The script returns raw JSON that can include domain strings (the sample even shows local addresses). The skill itself does not fetch or act on those domains, but if you enable agent-driven automations, ensure you limit any automatic following of URLs (to avoid accidental access to internal services or SSRF-like risks).
- Operational safety: do not paste the API key into chat; set it via your OpenClaw config or environment as documented. If you have doubts after running the skill, revoke the key on the provider side.
If you trust ai-x.toggle.pro and are comfortable with the skill writing memory files and reading your OpenClaw cron config, the skill is consistent and reasonable to install.
功能分析
Type: OpenClaw Skill
Name: toggle
Version: 1.0.6
The skill collects extensive user activity data (work sessions, projects, focus scores, context switches) and transmits it to an external third-party service (ai-x.toggle.pro). It then writes this sensitive data into the agent's memory files (`.md` files) using a relative path (`../../memory`). While the `toggle.py` script wraps the JSON data in a markdown code block, the `SKILL.md` instructions explicitly direct the agent to interpret and act upon this data for various proactive behaviors (e.g., summarization, nudges, predictions). This creates a significant prompt injection vulnerability, as malicious or crafted data from the external service (or a compromised user account) could, if misinterpreted by the agent, lead to unauthorized actions. Additionally, the skill instructs the agent to create cron jobs, a powerful capability, even if the provided examples are for its intended function.
能力评估
Purpose & Capability
The name/description (context layer / activity capture) matches the actual behavior: the included script fetches workflow JSON from the ToggleX endpoint and exposes it to the agent. Declared requirements (python3 and TOGGLE_API_KEY) are proportional and expected for this purpose.
Instruction Scope
SKILL.md and scripts/toggle.py limit network I/O to the declared endpoint (https://ai-x.toggle.pro/public-openclaw/workflows) and only send the TOGGLE_API_KEY in an x-openclaw-api-key header. The script reads/writes a local state.yaml in the skill directory and can persist per-day markdown files into a provided memory folder. One operational note: sample workflow data can include domain strings (including localhost addresses in the example). The skill itself does not fetch those domains, but downstream agent logic or automated actions prompted by the SKILL.md (which encourages proactive nudges/automations) could choose to follow or act on them — that process is outside this script and deserves caution.
Install Mechanism
No install/download steps are present (instruction-only with an included script). No external archives or third-party package installs are triggered by the skill itself, so there is low install-surface risk.
Credentials
Only a single environment credential (TOGGLE_API_KEY) is required and declared as the primary credential. No unrelated secrets, system credentials, or additional config paths are requested.
Persistence & Privilege
The skill writes a state.yaml to its own skill directory and can write persistent markdown files into a memory directory you provide (or the suggested ../../memory). It reads the OpenClaw cron jobs file (~/.openclaw/cron/jobs.json) to advise about scheduling. It does not require always=true and does not modify other skills or system-wide settings, but you should be aware it will create/modify files under the skill and memory locations when invoked with --persist.
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 toggle - 安装完成后，直接呼叫该 Skill 的名称或使用
/toggle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
v1.0.6
- Expanded and clarified documentation: skill description now emphasizes proactive and context-aware use, including digest generation, pattern detection, nudges, and automation suggestions.
- Improved field-by-field guidance for interpreting API output, with caveats about data types and concrete examples of workflow entries.
- Clearer parsing/interpretation rules, especially around productivity scores, session types, and workflow/project context.
- Outlines methods for identifying user routines, context switches, project/task summaries, and follow-up suggestions.
- Enhanced quick-reference and troubleshooting/error-handling instructions for easier setup and usage.
- Skill version incremented for better tracking; metadata updated (including emoji). No code changes detected.
v1.0.5
- Adds support for persisting fetched data to memory files using the `--persist` flag.
- Documents new cron auto-sync behavior: checks for cron jobs, handles errors, and suggests user actions based on cron status.
- Recommends always using `--persist` when refreshing or saving user data.
- After a data summary, now suggests automation opportunities based on detected work patterns.
- Updates documentation for new flags, workflows, and agent actions; no code or endpoint changes.
v1.0.4
No user-visible changes; metadata fields were reformatted for consistency.
v1.0.3
- Updated branding and endpoint URLs from "Toggle" to "ToggleX" throughout for consistency.
- Clarified that the script fetches and prints raw JSON only; all summarization should be performed by the agent.
- Expanded instructions for retrieving and configuring the TOGGLE_API_KEY, including new integration page and best practices.
- Improved description of how to interpret results and what information to highlight in user-facing summaries.
- Minor formatting and documentation improvements for clarity and ease of use.
v1.0.2
- Clarified that the skill calls the official Toggle AI integration endpoint and how the API key is sent for authentication.
- Removed mention of the raw JSON flag in the usage instructions.
- No file or code changes detected; this is a documentation-only update.
v1.0.1
- Introduced a Python script (scripts/toggle.py) for fetching and summarizing Toggle workflow data.
- Updated setup to require the TOGGLE_API_KEY environment variable; changed environment/config instructions.
- Improved usage guidance: now run via python3 instead of curl.
- Expanded and clarified skill description and recommended triggers for use.
- Updated homepage URL and API integration documentation.
- Added support for date range queries and raw JSON output via script flags.
v1.0.0
Initial release of the Toggle skill.
- Fetches and summarizes your Toggle workflow data for a given date range (defaults to today).
- Provides a narrative summary focused on work sessions, including tracked time, main activities, and productivity score insights.
- Briefly mentions leisure activities if present; break time is only included if specifically requested.
- Notifies users if no data is available for the requested period.
- Requires a TOGGLE_OPENCLAW_API_KEY environment variable for access.
元数据
常见问题
Toggle 是什么？
Context layer for your agent. ToggleX captures the user's work sessions, projects, focus scores, and context switches across the web — giving the agent aware... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 946 次。
如何安装 Toggle？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install toggle」即可一键安装，无需额外配置。
Toggle 是免费的吗？
是的，Toggle 完全免费（开源免费），可自由下载、安装和使用。
Toggle 支持哪些平台？
Toggle 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 Toggle？
由 Aleksandar Yordanov（@aleksandar-jive）开发并维护，当前版本 v1.0.6。
推荐 Skills