← Back to Skills Marketplace
3564
Downloads
7
Stars
32
Active Installs
4
Versions
Install in OpenClaw
/install token-manager
Description
Universal LLM Token Manager - Monitor usage and provide cost-saving recommendations for Kimi, OpenAI, Anthropic, Gemini, and local models. Features scheduled monitoring, cross-session tracking, and proactive alerts.
Usage Guidance
Review before installing. Prefer environment variables for API keys, do not pass secrets as command arguments, avoid sending sensitive prompt text through provider token-estimation APIs, and remove any cron job when you no longer want recurring balance checks.
Capability Analysis
Type: OpenClaw Skill
Name: token-manager
Version: 1.2.0
The OpenClaw AgentSkills skill bundle 'token-manager' is classified as benign. All code (`scripts/manager.js`, `scripts/scheduler.js`, `scripts/session-tracker.js`) and documentation (`SKILL.md`) align with the stated purpose of an LLM token manager. API keys are correctly handled via environment variables, data is stored locally in the `.data/` directory, and network requests are exclusively directed to official LLM API endpoints (e.g., api.moonshot.cn, api.openai.com). While the skill utilizes the `openclaw cron add` command for scheduled monitoring, this is a legitimate feature used for its declared purpose and does not indicate malicious persistence. No evidence of data exfiltration, malicious execution, obfuscation, or prompt injection attempts to subvert the agent's control or steal sensitive information was found.
Capability Assessment
Purpose & Capability
Token usage reporting, balance checks, local history, and optional scheduled alerts are coherent with the stated token-manager purpose.
Instruction Scope
The documentation and code contradict the security claim that API keys are read only from environment variables: documented commands and manager.js also accept API keys as positional CLI arguments.
Install Mechanism
No hidden installer or automatic setup was found; cron and tool registration are presented as manual user actions.
Credentials
Local .data storage and requests to provider APIs are proportionate for this purpose, but the security text understates that provider API calls may receive credentials and, through the exported token-estimation function, message text.
Persistence & Privilege
Persistent local state and scheduled cron checks are disclosed and purpose-aligned, though users should understand cron checks continue until disabled.
How to Use
- Make sure OpenClaw is installed (local or Docker)
- Run the install command in chat:
/install token-manager - After installation, invoke the skill by name or use
/token-manager - Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.2.0
**Summary: Adds scheduled monitoring and cross-session analytics with proactive alerts.**
- Introduced scheduled monitoring and balance alerts via new scripts (`scheduler.js`).
- Added cross-session tracking and analytics with session recording and reporting (`session-tracker.js`).
- Enhanced documentation: streamlined SKILL.md, added setup guides, new commands, and clear cost-saving recommendations.
- Provides sample cron job setups and tool integration steps.
- All monitoring/alert state and session data are stored locally for privacy.
- No changes to API usage or keys—still only local storage and official APIs.
v1.1.1
## Token Manager v1.1.1
- Added "安全说明 / Security" section to documentation, detailing privacy and API key handling practices.
- Clarified that all data stays local, API keys are only read from environment variables, and only official provider APIs are used.
- No functional or script changes outside documentation.
v1.1.0
**v1.1.0 introduces multi-provider support and enhanced features.**
- Added support for multiple LLM providers: Moonshot/Kimi, OpenAI, Anthropic/Claude, Google Gemini, and Ollama.
- New provider management commands: list, set default, and handle API keys.
- Universal balance query, cost estimation, and report functions now support provider and model selection.
- Automated detection of each provider’s capabilities (e.g., balance query, estimation accuracy).
- Unified cost-saving recommendations and pricing reference for all supported models.
v1.0.0
Token Manager 1.0.0 – Initial release
- Real-time monitoring and reporting for Kimi/Moonshot API token usage
- Balance query and cost estimation functionality
- Smart, context-based cost-saving suggestions (context compaction, sub-agent spawning, disabling reasoning)
- Usage history tracking
- Quick action recommendations for session optimization
Metadata
Frequently Asked Questions
What is Token Manager?
Universal LLM Token Manager - Monitor usage and provide cost-saving recommendations for Kimi, OpenAI, Anthropic, Gemini, and local models. Features scheduled monitoring, cross-session tracking, and proactive alerts. It is an AI Agent Skill for Claude Code / OpenClaw, with 3564 downloads so far.
How do I install Token Manager?
Run "/install token-manager" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.
Is Token Manager free?
Yes, Token Manager is completely free (open-source). You can download, install and use it at no cost.
Which platforms does Token Manager support?
Token Manager is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).
Who created Token Manager?
It is built and maintained by Felix (@kelegele); the current version is v1.2.0.
More Skills