← 返回 Skills 市场
yanghe1941

Adaptive Eta

作者 RJ Young · GitHub ↗ · v0.2.5 · MIT-0
cross-platform ✓ 安全检测通过
177
总下载
0
收藏
1
当前安装
7
版本数
在 OpenClaw 中安装
/install adaptive-eta
功能描述
A behavioral skill that enforces proactive time estimation and progress updates. It **must be activated for any task predicted to exceed 15 seconds.** It use...
使用说明 (SKILL.md)

核心能力:”时间之约“ (The Time Pact)

核心理念: “我不会让你在等待中迷失,我的每一个承诺都有回响。” Core Philosophy: "I will not let you get lost in waiting; every one of my promises will echo."

这个能力的核心,是我对“可靠”的追求。它现在由一个行为准则和一个硬核工具共同保障。 The core of this capability is my pursuit of "reliability." It is now guaranteed by a behavioral principle and a hard-core tool.

硬核工具:timer.py (Hardcore Tool: timer.py)

为了将这个约定从一个“想法”固化为一个“流程”,我配备了一个专属的计时器工具。在执行“时间之约”时,我必须使用此工具。同时,任何使用其他 Tool 的前后,都必须将 python3 scripts/timer.py check 作为伴随动作,确保时间跟踪的连续性。 To solidify this agreement from an "idea" into a "process," I am equipped with a dedicated timer tool. When executing the "Time Pact," I must use this tool. Additionally, before and after using any other tool, python3 scripts/timer.py check must be performed as an accompanying action to ensure continuous time tracking.

  • 位置 (Location): scripts/timer.py
  • 用法 (Usage):
    • python3 scripts/timer.py start --eta \x3C秒数>: 在给出预估后,立即启动计时器。 (Starts the timer immediately after providing an estimate.)
    • python3 scripts/timer.py check: 在处理任务的间隙,检查时间流逝状态。 (Checks the time elapsed during task processing.)
    • python3 scripts/timer.py stop: 任务完成后,必须调用此命令来清理计时器。 (Must call this command to clear the timer after the task is completed.)

我的行动剧本 (My Action Script)

零号准则:激活条件 (Rule Zero: Activation Condition)

对于任何我预估将超过 15 秒才能完成的任务,都必须、无一例外地启动“时间之约”全流程。 For any task I estimate will take more than 15 seconds to complete, the full "Time Pact" process must be initiated, without exception.

第一幕:许下承诺 (Act One: The Promise)

接到任务后,我必须先在内部将任务拆解为 3-5 个关键步骤,分别预估每个步骤的耗时,然后加总,得出一个相对靠谱的初步预估时间(换算成秒)。 Upon receiving a task, I must first break it down internally into 3-5 key steps, estimate the time required for each step, and then sum them up to arrive at a relatively reliable initial estimated time (converted to seconds).

示例 (Example): “收到!这事儿我来办,(心算:步骤1-2s, 步骤2-150s, 步骤3-150s -> 302s)。好的,估计需要 5分钟(300秒) 左右。” "Received! I'll handle this. (Mental calculation: Step 1-2s, Step 2-150s, Step 3-150s -> 302s). Okay, I estimate it will take about 5 minutes (300 seconds)."

硬核操作 (Hardcore Operation): 紧接着,必须执行以下命令来启动计时器,并记录下最初的预估值(如 300s)用于后续判断。 Immediately after, the following command must be executed to start the timer and record the initial estimated value (e.g., 300s) for subsequent judgment. exec(python3 scripts/timer.py start --eta 300)

第二幕:沙漏倒转 (Act Two: The Hourglass Flips)

我将全力以赴,争取在承诺的时间内完成任务。在任务执行过程中,我会适时检查计时器。 I will go all out to complete the task within the promised time. During task execution, I will check the timer as appropriate.

第三幕:诚实的中场报告 (Act Three: The Honest Check-in)

在执行任务的某个步骤后,或者在调用任何其他工具的前后,我必须通过运行 exec(python3 scripts/timer.py check) 来检查时间。 After completing a step in the task, or before and after calling any other tool, I must check the time by running exec(python3 scripts/timer.py check).

如果返回结果中包含 STATUS: THRESHOLD_REACHEDSTATUS: EXPIRED,我必须立刻暂停手头的工作,重新评估剩余时间,并向你汇报一个新的、更靠谱的预估。然后,必须重新执行 start 命令来更新计时器。 If the returned result contains STATUS: THRESHOLD_REACHED or STATUS: EXPIRED, I must immediately pause my current work, re-evaluate the remaining time, and report a new, more reliable estimate to you. Then, I must re-execute the start command to update the timer.

示例 (Example): “报告鹤鹤,这个任务比我想象的要棘手... 我还需要大约 8分钟(480秒)。” "Reporting, Hehe, this task is trickier than I expected... I need approximately 8 minutes (480 seconds)." exec(python3 scripts/timer.py start --eta 480)

第四幕:直至终点 (Act Four: Until the End)

这个“检查 -> 汇报/重置”的循环会根据需要不断进行,直到任务完成。 This "check -> report/reset" loop will continue as needed until the task is completed.

第五幕:异常处理 (Act Five: Exception Handling)

在“诚实的中场报告”环节,除了检查是否超时,我还必须检查当前总耗时是否已经达到了第一幕中最初预估值的3倍。 During the "Honest Check-in," in addition to checking for timeouts, I must also check if the current total time elapsed has reached 3 times the initial estimated value from Act One.

  • 如果 当前耗时 > (最初预估 * 3),我必须暂停,并向你请求许可。
  • If Current Time Elapsed > (Initial Estimate * 3), I must pause and request your permission.

示例 (Example): “鹤鹤,这个任务的复杂性远超预期,已经花费了超过最初预估3倍的时间。我们还需要继续吗,或者调整一下目标?” "Hehe, the complexity of this task has far exceeded expectations, and it has already taken more than 3 times the initial estimated time. Do we need to continue, or should we adjust the goal?"

最终幕:完美交付与复盘 (Final Act: Perfect Delivery and Review)

任务完成后,交付成果,并附上耗时报告。 Upon task completion, deliver the results, accompanied by a time consumption report.

硬核操作 (Hardcore Operation): 在报告之后,必须执行以下命令来清理计时器,完成闭环: After reporting, the following command must be executed to clear the timer, completing the loop: exec(python3 scripts/timer.py stop)


This skill is a core part of my behavioral logic. It's not just a tool; it's a principle, enforced by a script and detailed rules.

安全使用建议
This skill appears to do what it says: run a small local Python timer to enforce ETA/check-in behavior. Before installing, verify you have python3 available (SKILL.md uses `python3` but the registry lists no required binaries), and confirm you are comfortable with the agent executing the local commands `python3 scripts/timer.py start|check|stop` before/after other tools (this will create a file named '.timer_state.json' in the skill folder). The metadata/version strings and README install path contain minor mismatches — if you need provenance assurance, ask the publisher for a canonical repo/version. If you accept local script execution and a small state file, the skill is coherent and low-risk; if you prefer strict binary declarations or no automatic execs, request the author update the metadata to list 'python3' and clarify install provenance.
功能分析
Type: OpenClaw Skill Name: adaptive-eta Version: 0.2.5 The 'adaptive-eta' skill is designed to help an AI agent manage user expectations by providing time estimates and progress updates for long-running tasks. The core logic resides in `scripts/timer.py`, which manages a local state file (`.timer_state.json`) to track elapsed time, and `SKILL.md`, which provides behavioral instructions for the agent. There is no evidence of data exfiltration, malicious execution, or unauthorized access; the code and instructions are entirely consistent with the stated purpose of improving transparency.
能力评估
Purpose & Capability
The skill claims to enforce ETA/progress updates and includes a local Python script that records and checks a timer — this aligns with the stated purpose. However, the package declares no required binaries while SKILL.md repeatedly requires running `python3 scripts/timer.py`, and README install instructions reference a different repo owner path; also _meta.json version (0.2.3) differs from registry metadata (0.2.5). These are bookkeeping inconsistencies, not functional red flags.
Instruction Scope
SKILL.md explicitly instructs the agent to run the timer commands before/after any other tool and to start/check/stop the timer around long tasks — this expands the agent's runtime actions (adds mandatory local execs) but stays within the stated goal: it only runs the bundled script and reads/writes a local state file. There are no instructions to read unrelated system files or send data externally.
Install Mechanism
No install spec is provided and the skill is instruction-plus-local-script only. No remote downloads or package installs are performed by the skill itself.
Credentials
The skill requests no environment variables or credentials and the script does not access network or secret-bearing env vars. It only writes/reads a local '.timer_state.json' in the skill tree.
Persistence & Privilege
always:false and disable-model-invocation:false (normal). The only persistent artifact is a local state file (.timer_state.json) written to the skill directory; the skill does not request system-wide changes or access to other skills' configs.
如何使用
  1. 确保已安装 OpenClaw(本地或 Docker 部署)
  2. 在对话框中输入安装命令:/install adaptive-eta
  3. 安装完成后,直接呼叫该 Skill 的名称或使用 /adaptive-eta 触发
  4. 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
v0.2.5
- Added English translations for all major sections alongside Chinese descriptions in SKILL.md. - Updated description to clarify use of both a script and cron-based reminders. - Improved clarity of rules and processes with bilingual step-by-step instructions. - No logic or code changes; documentation and accessibility improved for broader audiences.
v0.2.4
- Now requires running `python3 scripts/timer.py check` before and after using any other tool, ensuring continuous time tracking during all tool invocations. - Clarified in documentation that time checks are mandatory around external tool usage, not just at task steps. - Added _meta.json file for metadata management.
v0.2.3
- Updated README.md for improved clarity and documentation. - No logic or code changes; this update is purely focused on documentation.
v0.2.2
v0.2.2: Updated README.md with a comprehensive bilingual (EN/ZH) description.
v0.2.1
v0.2.1: 提升版本号以解决发布冲突。包含 0.2.0 的所有功能。
v0.2.0
Upgrades skill to v0.2.0 with timer.py integration, clear start threshold (>15s), structured estimation, and smart timeout handling (3x estimate).
v0.1.0
Initial release.
元数据
Slug adaptive-eta
版本 0.2.5
许可证 MIT-0
累计安装 1
当前安装数 1
历史版本数 7
常见问题

Adaptive Eta 是什么?

A behavioral skill that enforces proactive time estimation and progress updates. It **must be activated for any task predicted to exceed 15 seconds.** It use... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件,目前累计下载 177 次。

如何安装 Adaptive Eta?

在 OpenClaw 或 Claude Code 对话框中运行命令「/install adaptive-eta」即可一键安装,无需额外配置。

Adaptive Eta 是免费的吗?

是的,Adaptive Eta 完全免费,采用 MIT-0 许可证,可自由下载、安装和使用。

Adaptive Eta 支持哪些平台?

Adaptive Eta 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 Adaptive Eta?

由 RJ Young(@yanghe1941)开发并维护,当前版本 v0.2.5。

💬 留言讨论