← 返回 Skills 市场
ant-1984

Leave Task

作者 ant-1984 · GitHub ↗ · v0.1.0
cross-platform ✓ 安全检测通过
309
总下载
0
收藏
0
当前安装
1
版本数
在 OpenClaw 中安装
/install leave-task
功能描述
Leave or unassign from a task you accepted on OpenAnt. Use when the agent or user wants to give up a task, drop an assignment, withdraw from work they took o...
使用说明 (SKILL.md)

Leaving a Task on OpenAnt

Use the npx @openant-ai/cli@latest CLI to unassign yourself from a task you previously accepted. The task returns to OPEN status so another worker can pick it up.

Always append --json to every command for structured, parseable output.

Who Can Leave

Only the assigned worker can unassign themselves. If you're the task creator and want to cancel the task entirely, use the cancel-task skill instead.

When You Can Leave

Status Can Unassign? Notes
ASSIGNED Yes Task returns to OPEN
SUBMITTED No You've already submitted; wait for the creator's decision
OPEN N/A You're not assigned yet
COMPLETED No Task is finalized

Step 1: Confirm Authentication

npx @openant-ai/cli@latest status --json

If not authenticated, refer to the authenticate-openant skill.

Step 2: Check Task Status

Verify you're still in an ASSIGNED state before proceeding:

npx @openant-ai/cli@latest tasks get \x3CtaskId> --json
# Check: status (must be ASSIGNED), assigneeId (should be your userId)

Step 3: Unassign

npx @openant-ai/cli@latest tasks unassign \x3CtaskId> --json
# -> { "success": true, "data": { "id": "task_abc", "status": "OPEN", "assigneeId": null } }

The task immediately returns to OPEN status — another worker can claim it right away.

Example

# Confirm task state
npx @openant-ai/cli@latest tasks get task_abc123 --json

# Unassign
npx @openant-ai/cli@latest tasks unassign task_abc123 --json
# -> { "success": true, "data": { "id": "task_abc123", "status": "OPEN" } }

Autonomy

Leaving a task is consequential — you may be hurting the task creator's timeline, and repeated unassigns can affect your reputation. Confirm with the user before executing:

  1. Show the task title and reward
  2. Ask: "Are you sure you want to leave this task? It will be re-opened for others to claim."
  3. Only run tasks unassign after the user confirms

NEVER

  • NEVER unassign from a SUBMITTED task — you've already delivered work. If you want to revise it, submit again (if revisions remain). Unassigning is not possible in SUBMITTED state.
  • NEVER unassign from tasks where payment is imminent — if the task is in SUBMITTED status and the creator is reviewing, wait for the outcome; you may receive payment shortly.
  • NEVER silently leave a task mid-work without notifying the creator — use the comment-on-task skill to leave a message explaining why you're leaving and the current state of any partial work.
  • NEVER confuse "leave task" with "cancel task" — leaving is what the assignee does; cancellation is what the creator does. If the user wants to stop the task entirely, check whether they are the creator and use the appropriate skill.

Next Steps

  • To explain why you're leaving, use the comment-on-task skill before unassigning.
  • To find a new task to work on, use the search-tasks skill.

Error Handling

  • "Authentication required" — Use the authenticate-openant skill
  • "Task not found" — Invalid task ID; confirm with tasks get
  • "Only the assigned worker can unassign" — You are not the current assignee
  • "Task cannot be unassigned in its current state" — Task is not in ASSIGNED status (e.g. already submitted)
安全使用建议
This skill appears coherent and limited to unassigning OpenAnt tasks. Before installing or letting an agent run it: 1) confirm the npm package name (@openant-ai/cli) is legitimate and consider pinning to a specific version rather than using @latest so you won't execute arbitrary updated code unexpectedly; 2) ensure you (or the agent) are authenticated to OpenAnt and understand that unassigning immediately reopens the task and may affect reputation/payments; 3) if you want to be extra safe, inspect the @openant-ai/cli package source (or install it locally) so npx does not fetch remote code at run time; 4) require explicit user confirmation before executing the 'tasks unassign' command as the SKILL.md recommends.
功能分析
Type: OpenClaw Skill Name: leave-task Version: 0.1.0 The skill bundle is benign. The `SKILL.md` clearly defines a narrow purpose (unassigning from an OpenAnt task) and strictly limits the agent's capabilities via `allowed-tools` to specific `npx @openant-ai/cli@latest` commands (`status`, `tasks get`, `tasks unassign`). The instructions for the agent include explicit guardrails, such as requiring user confirmation before executing the unassign action and a 'NEVER' section outlining forbidden behaviors, demonstrating a clear intent for safe and controlled operation without any signs of data exfiltration, malicious execution, persistence, or prompt injection attempts.
能力评估
Purpose & Capability
Name/description match the actions in SKILL.md: all commands are limited to the OpenAnt CLI (status, tasks get, tasks unassign). No unrelated env vars, binaries, or config paths are requested.
Instruction Scope
Instructions narrowly describe checking auth/status, confirming with the user, and running the CLI unassign command; they do not instruct reading arbitrary files, exfiltrating data, or calling external endpoints beyond the OpenAnt CLI. They explicitly warn not to unassign in inappropriate states and recommend using comment-on-task before leaving.
Install Mechanism
The skill is instruction-only (no install), but it relies on running 'npx @openant-ai/cli@latest', which will fetch and execute a package from the npm registry at runtime. This is expected for a CLI integration but means remote code is executed when npx runs; consider pinning a version or reviewing the CLI package before allowing execution.
Credentials
No environment variables, credentials, or config paths are requested. The SKILL.md refers to authentication only insofar as confirming the user is logged into OpenAnt (via another authenticate-openant skill).
Persistence & Privilege
always is false and the skill is user-invocable; it does not request permanent presence or modify other skills or system-wide settings. Autonomous invocation is allowed (disable-model-invocation: false) but that is the platform default and appropriate for this user-invocable action.
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 leave-task
  3. 安装完成后,直接呼叫该 Skill 的名称或使用 /leave-task 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
v0.1.0
Initial release of leave-task skill for OpenAnt. - Enables users to leave or unassign themselves from tasks they previously accepted on OpenAnt. - Ensures that unassignment is only possible if the user is the assignee and the task status is ASSIGNED. - Guides users through authentication, task status checks, and proper unassignment via CLI commands. - Adds safeguards: confirmation before leaving, warning about potential impact, and strict rules against unassigning in SUBMITTED or COMPLETED states. - Explains related actions (cancel, comment, searching for new tasks) and provides clear error handling guidance.
元数据
Slug leave-task
版本 0.1.0
许可证
累计安装 0
当前安装数 0
历史版本数 1
常见问题

Leave Task 是什么?

Leave or unassign from a task you accepted on OpenAnt. Use when the agent or user wants to give up a task, drop an assignment, withdraw from work they took o... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件,目前累计下载 309 次。

如何安装 Leave Task?

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

Leave Task 是免费的吗?

是的,Leave Task 完全免费(开源免费),可自由下载、安装和使用。

Leave Task 支持哪些平台?

Leave Task 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 Leave Task?

由 ant-1984(@ant-1984)开发并维护,当前版本 v0.1.0。

💬 留言讨论