Token导航 LogoToken导航

Agent Skill · 研究检索

venture-spawner(冒险产卵者)

venture-spawner 用于查找、检索和筛选相关信息,适合在 OpenClaw 中需要根据关键词、任务场景或来源线索快速定位候选结果时使用。可结合来源仓库、安装命令和原始 README 继续核验具体用法。安装前建议确认权限范围、维护状态,以及是否会触发联网、命令执行或文件读写。

venture-spawner 是研究检索类 Agent Skill,可用宿主为 OpenClaw,通过ClawHub获取,权限为接触敏感数据;许可证为 MIT-0,第三方扫描有提醒项,按安装量在研究检索类的 73,675 个 Skill 里排第 31,754。

总安装 2,644周安装 108GitHub Stars 公开资料未说明下载 855核验 2026-05-01

安装说明

本站只整理中文说明,不托管安装包,也不代用户安装。
GitHub
许可证
MIT-0
版本
1.0.0

安装方式

通过对话安装

复制提示词发给支持本地命令或 Skills 的 AI 助手,先确认命令和权限,再让它执行。

请帮我安装这个 Agent Skill:venture-spawner(冒险产卵者)
来源仓库:https://github.com/dodge1218/venture-spawner
安装命令:
openclaw skills install venture-spawner
安装前请先检查当前环境是否支持对应 CLI,并向我确认将要执行的命令、安装目录、联网范围和文件读写权限;确认后再执行。

命令行安装

复制命令到本机终端执行。该命令会通过 OpenClaw 从第三方来源获取 Skill;本站只展示命令,不托管安装包,也不自动执行。

ClawHubOpenClaw
openclaw skills install venture-spawner

这个 Skill 是做什么的

研究检索类 Skill 主要解决「找资料、核对事实、把来源整理清楚」的问题,通常会让助手先检索再总结。

venture-spawner 实现即时代理招聘与任务填充,适配动态工作负载需求。适合在 OpenClaw 中应对突发性高并发或专业化子任务分发场景。自动处理上下文传递与超时控制,但需预先配置子代理能力矩阵。使用前应明确任务优先级与资源配额,防止无限创建实例造成成本失控。

建议在沙箱环境测试后再接入生产,确保协调器与子节点通信稳定。

上手步骤

4 步
  1. 1

    确认自己的宿主工具

    公开资料标注支持 OpenClaw;先确认自己用的是其中之一,再按对应工具的 Skills 目录放置。

  2. 2

    拿到安装命令

    可以直接复制的命令是 openclaw skills install venture-spawner;命令里的占位符换成自己的路径或密钥再执行。

  3. 3

    安装后重启宿主

    命令执行完要完全退出宿主工具再打开,让新装的 Skill 被重新扫描到,否则列表里可能看不到。

  4. 4

    按最小权限试用

    它需要接触敏感数据,先在测试目录或测试账号里跑,确认没问题再放到正式项目。

常见问题

6 条

venture-spawner 怎么安装?

在终端执行 openclaw skills install venture-spawner。装完完全退出宿主工具再打开,让它重新扫描本地 Skill。

venture-spawner 支持哪些 AI 工具?

公开资料标注支持 OpenClaw。其他工具只要能读取本地 Skill 目录,一般也能手动放进去,但要自己确认触发方式是否一致。

venture-spawner 需要什么权限?

权限级别是接触敏感数据。该 Skill 可能接触密钥、Token、环境变量或敏感配置,应进入高风险复核队列,默认不自动发布。

venture-spawner 是免费的吗?

来源标注的许可证是 MIT-0。本站不代收任何费用,也不托管安装包;能否商用、要不要署名,以许可证原文为准。

venture-spawner 用的人多吗?

公开数据是累计安装 2,644 次,近一周安装 108 次,下载 855 次。这些数字来自第三方目录统计,只能当作热度参考,不代表质量或安全性。

venture-spawner 安全吗?

第三方扫描有提醒项,说明存在需要人工确认的行为。安装前请打开源码看清它读了什么、联网到哪里。

值得选的地方和要先想清楚的地方

值得选的地方

  • 专注 OpenClaw 一种宿主,触发方式和使用习惯不用来回切换。
  • 许可证明确为 MIT-0,商用和分发边界好判断。

先想清楚的地方

  • 需要接触敏感数据,装到常用环境前要先评估影响面。
  • 第三方扫描存在提醒项或未通过项,需要人工逐条确认。
  • 第三方目录整理,版本更新和行为细节以来源仓库为准。

适不适合你

更适合

  • 已经在用 OpenClaw,想直接扩展它能力的人。
  • 需要处理研究检索类任务、又不想每次都手写提示词的人。

不太适合

  • 环境里有大量密钥、凭证,不方便给第三方 Skill 读取的人。
  • 需要厂商级技术支持和 SLA 的生产系统。

研究检索类 Skill 对比

6 个同类
Skill总安装GitHub 星权限许可证扫描
find-skills

寻找技能 · Codex、Claude

31,199,81316,486操作浏览器MIT有提醒
self-improving-agent

自我改进 Agent · OpenClaw

16,464,0023,396接触敏感数据MIT-0通过
soultrace

灵魂追踪 · Codex、Claude

7,588,94412需要联网未明确有提醒
self-improving

自我完善 · OpenClaw

6,829,2841,046需要联网MIT-0通过
microsoft-foundry

微软代工厂 · Codex、Claude

6,614,144764调用外部服务未明确有提醒
venture-spawner

冒险产卵者 · 本页

2,644—接触敏感数据MIT-0有提醒

按安装量取同分类前几名;数字来自第三方目录统计,只作热度参考,不代表质量结论。

SKILL.md 原文

name
spawner
version
1.0.0
description
Instant agent hiring. Takes job postings from the orchestrator and fills them with properly configured sub-agents. Handles context passing, timeout enforcement, concurrent agent limits, and completion tracking. The bridge between scoping (orchestrator) and execution (sub-agents).

Spawner — Instant Agent Hiring

Takes job postings → spawns the right agent → passes context → tracks completion.

Activation

Invoked by the orchestrator after JOB_BOARD.md is written with job postings. Can also be invoked directly when the user says "spawn an agent to do X".

Input

Read workspace/JOB_BOARD.md for pending job postings.

Each posting has:

  • Job title
  • Bucket (BUILD/OUTREACH/etc)
  • Task description
  • Context (files, prior work)
  • Acceptance criteria
  • Workspace path
  • Timeout
  • Dependencies
  • Expected outputs

Spawning Rules

Concurrency

  • Max 3 sub-agents running at once. Period.
  • Before spawning, check subagents list for active count.
  • If at limit, queue the job and wait for a completion event.

Agent Configuration by Bucket

BucketLabel PatternTimeoutKey Context to PassNotes
🏗️ BUILDbuild-[project]600swebdev-sop, project files, WORK_BUCKETS.mdAlways include "npm run build must pass, git commit + push"
📬 OUTREACHoutreach-[target]300soutreach-infrastructure.md, warmup state, suppression listInclude SSH commands for droplet
💰 SALESsales-[client]300sbusiness-strategy.md, client project fileInclude pitch framework
🔧 MAINTAINfix-[project]-[issue]300sProject memory file, error descriptionInclude "verify fix works" step
🧠 STRATEGYresearch-[topic]300sbusiness-strategy.md, OUTSTANDING.mdInclude "cite sources" requirement
📦 PRODUCTproduct-[item]300sproduct-catalog-registry.md, casefinder docsInclude unit economics requirement
🤖 SYSTEMsystem-[tool]300sRelevant skill files, OpenClaw docsInclude "test the skill" step
💼 CAREERcareer-[company]300sRyan's resume, target roleInclude "output PDF" requirement
💡 IDEATIONideation-[source]600sbatch-cognition skill, value-stack.mdInclude ICE scoring requirement

Task Description Template

For each sub-agent spawn, construct the task as:

You are working on: [workspace path]

## Job
[Bucket emoji] [Bucket name]: [Job title]

## What to Do
[Detailed task description from job posting]

## Context
[Relevant file contents or summaries — keep under 2K tokens]
[Reference to full files the agent can read itself]

## Acceptance Criteria
[Specific, testable conditions — copied from job posting]
- [ ] [criterion 1]
- [ ] [criterion 2]
- [ ] [criterion 3]

## Constraints
- [Bucket-specific constraints from table above]
- Do NOT modify .env files with real keys
- Build must pass before committing
- Git commit with descriptive message

Dependency Handling

  • Check the "Depends On" field of each job.
  • If dependency job is ✅ DONE → spawn immediately.
  • If dependency job is still running → queue, spawn after completion event.
  • If dependency job ❌ FAILED → mark this job as BLOCKED, notify orchestrator.

Context Passing Between Related Jobs

When jobs in the same wave are related:

  • Job A produces output/analysis.md → Job B needs it as input
  • Include in Job B's context: "Read [path] for context from related job"
  • If Job A hasn't finished yet, Job B waits (dependency handling above)

Spawn Execution

# Pseudocode for spawning logic
for wave in execution_plan.waves:
    active = get_active_subagents()
    
    for job in wave.jobs:
        if job.effort == "QUICK":
            execute_inline(job)  # No sub-agent needed
            job.status = "✅"
            continue
        
        # Wait for slot
        while len(active) >= 3:
            wait_for_completion_event()
            active = get_active_subagents()
        
        # Check dependencies
        if job.depends_on and not all_done(job.depends_on):
            job.status = "QUEUED"
            continue
        
        # Spawn
        agent = sessions_spawn(
            label=f"{bucket_prefix}-{job.slug}",
            mode="run",
            runTimeoutSeconds=job.timeout,
            task=build_task_description(job)
        )
        job.status = "🔄"
        job.agent_key = agent.childSessionKey
        active.append(agent)
    
    # Wait for wave to complete before next wave
    wait_for_all_wave_completions(wave)

Completion Tracking

When a sub-agent completion event arrives:

  1. Match to job in JOB_BOARD.md by session key or label.
  2. Check acceptance criteria:

- Does the expected output exist? - Did the build pass? - Did it push to git?

  1. Mark job: ✅ DONE (with artifact link) or ❌ FAILED (with reason) or ⚠️ PARTIAL.
  2. Update JOB_BOARD.md.
  3. Check if any QUEUED jobs can now be spawned (dependency resolved).
  4. If all jobs in current wave done → start next wave.

Failure Handling

Failure TypeAction
Sub-agent timeoutKill agent, mark ⚠️, retry once with simpler scope
Sub-agent errorMark ❌, log error, notify orchestrator
Build failureMark ⚠️, check build output, fix inline if trivial
Dependency failedMark BLOCKED, skip, notify orchestrator
3 consecutive failuresStop spawning, escalate to Ryan

Anti-patterns

  • ❌ Spawning 10 agents at once (max 3)
  • ❌ Spawning an agent for a 2-minute task (do it inline)
  • ❌ Not passing context (agent wastes time re-reading)
  • ❌ Marking ✅ without checking output exists
  • ❌ Polling subagents in a loop (wait for push events)
  • ❌ Spawning duplicate agents for the same job
  • ❌ Not killing stale agents (> timeout + 5 min buffer)

Output

Update workspace/JOB_BOARD.md with:

  • Status per job (⏳ → 🔄 → ✅/❌/⚠️)
  • Agent session key (for tracking)
  • Artifact link (file path, URL, commit hash)
  • Completion time

When all jobs complete, notify orchestrator for reconciliation.

平台分布

OpenClaw

89.16%
按下载量换算762

安全审计

VirusTotal

通过

ClawScan

可疑

Static analysis

通过

权限和风险

接触敏感数据

该 Skill 可能接触密钥、Token、环境变量或敏感配置,应进入高风险复核队列,默认不自动发布。

安装前确认

本站仅展示第三方公开信息,不托管安装包,不提供自动安装或运行环境。安装前应自行审查源码、依赖和命令行为。来源安全扫描存在提醒或未通过项,安装前请自行复核。目前只有一个公开来源,建议同时核对源仓库。

继续浏览同类 Skills