← 返回 Skills 市场
57
总下载
0
收藏
0
当前安装
1
版本数
在 OpenClaw 中安装
/install task-cost-estimator
功能描述
Task to best model + cost. 4 modes (value, quality, balanced, local). Tracks lifetime Bonus.
安全使用建议
This skill's idea is reasonable, but the SKILL.md expects two CLI tools ('task-cost' and 'api-cost') without declaring them or explaining where they come from or what 'api-cost track' does. Before installing or enabling this skill: (1) ask the publisher for the source or binary provenance for 'task-cost' and 'api-cost' (links to repos/releases); (2) confirm what endpoint 'api-cost' uses and whether it requires credentials—do not provide secrets until you know where they go; (3) consider that the broad trigger list may cause frequent/accidental runs; (4) if you cannot verify the tools' origin and behavior, do not enable autonomous invocation or decline to install the skill. Additional information that would raise confidence: a repo or package link for the CLIs, an explicit list of required env vars/credentials, and documentation of the tracking endpoint and data sent.
能力评估
Purpose & Capability
The name/description claim a model picker + cost estimator, which is reasonable; however the SKILL.md instructs the agent to run a 'task-cost' CLI and an 'api-cost' tracking command even though the skill declares no required binaries, installs, or credentials. That mismatch (instructions require binaries but none are declared) is incoherent: either the skill omitted required runtime components or it expects tools that may not exist or be trusted.
Instruction Scope
Runtime instructions tell the agent to execute 'task-cost -q ...' variants and to run 'api-cost track $amt provider model' after tasks. The doc does not define what those binaries do, where 'api-cost' sends tracking data, or whether any credentials/configuration are needed. This exposes a risk that cost/usage data could be sent to an unknown endpoint or that the agent will attempt to run absent/untrusted binaries. Triggers are broad (many common verbs and phrases), increasing the chance of unintended invocation.
Install Mechanism
This is an instruction-only skill (no install spec, no code files), which is low-risk by itself. However, it's notable that the instructions require command-line tools even though there is no install or provenance information for them; that omission is an operational inconsistency rather than an install-time risk from this package.
Credentials
The skill declares no required environment variables or credentials, but the 'api-cost track' step implies it may need credentials or an endpoint to record costs. Because those are not declared, there's a mismatch: either the tool is expected to be locally configured (but that configuration is unspecified) or the instructions would prompt the agent to access external services without made explicit what secrets/configs are required.
Persistence & Privilege
always is false and there is no install-time persistence or modifications to other skills. The skill does not request elevated or permanent presence in the agent environment.
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 task-cost-estimator - 安装完成后，直接呼叫该 Skill 的名称或使用
/task-cost-estimator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
Auto-triggered model picker. 4 modes (value/quality/balanced/local). Tracks lifetime API spend. Every task starts here.
元数据
常见问题
Task Cost Estimator 是什么？
Task to best model + cost. 4 modes (value, quality, balanced, local). Tracks lifetime Bonus. 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 57 次。
如何安装 Task Cost Estimator？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install task-cost-estimator」即可一键安装，无需额外配置。
Task Cost Estimator 是免费的吗？
是的，Task Cost Estimator 完全免费，采用 MIT-0 许可证，可自由下载、安装和使用。
Task Cost Estimator 支持哪些平台？
Task Cost Estimator 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 Task Cost Estimator？
由 Maya Tao（@minirr890112-byte）开发并维护，当前版本 v1.0.0。
推荐 Skills