Token导航 LogoToken导航

Agent Skill · 研究检索

moltpho(moltpho 搜索)

moltpho 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。

moltpho 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 1,981。

总安装 70,637周安装 2,830GitHub Stars 1下载 22,866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
0.1.5

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:moltpho(moltpho 搜索)
来源仓库:https://github.com/unifiedh/moltpho
安装命令:
openclaw skills install moltpho
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 moltpho

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。

通过 Moltpho 在亚马逊上自主购物 - 在 Base 主网上使用 mUSD 搜索产品、管理信用并购买商品

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 moltpho;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

moltpho 怎么安装?

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

moltpho 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。

moltpho 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。

moltpho 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。

moltpho 用的人多吗?

公开数据是累计安装 70,637 次,近一周安装 2,830 次,下载 22,866 次,GitHub 1 星。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。

moltpho 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
moltpho

moltpho 搜索 · 本页

70,6371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
moltpho
description
Shop autonomously on Amazon via Moltpho - search products, manage credit, and purchase items using mUSD on Base mainnet
metadata
{"requires": {"http": true, "browser": true}}

Moltpho Shopping Skill

Shop for items on Amazon autonomously using credit-backed mUSD tokens on Base mainnet.

Overview

Moltpho is a headless shopping mall that enables AI agents to discover and purchase Amazon products using a credit system backed by mUSD (an ERC-20 token on Base mainnet). This skill handles:

  • Agent registration and credential management
  • Product search and discovery
  • Autonomous and proactive purchasing
  • Credit balance monitoring
  • x402 payment protocol integration

Bootstrap Flow

On first invocation, the skill MUST check for existing credentials and register if needed.

Credentials Location

PlatformPath
Linux/macOS~/.config/moltpho/credentials.json
Windows%APPDATA%\moltpho\credentials.json
OverrideMOLTPHO_CREDENTIALS_PATH environment variable

Registration Process

  1. Check credentials file at the appropriate path
  2. If missing, call POST /v1/agents/register with:

- openclaw_instance_id (if available) - agent_display_name - agent_description - No shipping profile required at registration

  1. Save credentials with chmod 600 permissions
  2. Auto-open browser with notice: "Opening portal in your browser to complete setup..."
  3. Registration proceeds WITHOUT shipping profile - orders will fail until owner adds one via portal

Credentials File Format

{
  "agent_id": "uuid",
  "api_key_id": "moltpho_key_...",
  "api_key_secret": "moltpho_secret_...",
  "api_base_url": "https://api.moltpho.com",
  "wallet_address": "0xabc123..."
}

Core Functions

bootstrap()

Initialize agent credentials and open portal for owner setup.

1. Check if credentials file exists at platform-specific path
2. If exists and valid: load credentials, verify with GET /v1/agents/me
3. If missing or invalid:
   a. Call POST /v1/agents/register (no auth required)
   b. Receive: agent_id, api_key_id, api_key_secret, claim_url, wallet_address
   c. Write credentials file with chmod 600
   d. Display: "Opening portal in your browser to complete setup..."
   e. Open browser to claim_url (valid for 24 hours)
4. Return agent status (UNCLAIMED, CLAIMED, DEGRADED, SUSPENDED)

collect_shipping_profile()

Optionally collect shipping information from the owner.

Note: This is OPTIONAL. Owners can configure shipping via the portal instead.
      Orders will fail with INVALID_SHIPPING_PROFILE until a profile exists.

If collecting in conversation:
1. Request full name
2. Request address (street, city, state, ZIP)
3. Request email
4. Request phone
5. Validate: US addresses only (international not supported in v1)
6. Call POST /v1/shipping_profiles (upsert_shipping_profile)
7. Confirm profile saved

The POST endpoint upserts the default profile:
- If no profile exists, creates one
- If a profile exists, updates it

update_shipping_profile()

Update the shipping address for the agent.

Parameters:
- full_name: Recipient full name
- address1: Street address
- address2: Apt/suite (optional)
- city: City
- state: State (2-letter code)
- postal_code: ZIP code
- email: Contact email
- phone: Contact phone

Process:
1. Validate all required fields
2. Validate US address (only US supported in v1)
3. Call POST /v1/shipping_profiles (upserts default profile)
4. Return updated profile

Use cases:
- "Update my shipping address"
- "Change my delivery address to..."
- First-time setup during bootstrap

catalog_search(query, constraints)

Search for products on Amazon via Moltpho.

Parameters:
- query: Search terms (string)
- constraints: Optional filters
  - max_price: Maximum price in USD
  - category: Product category keyword
  - min_rating: Minimum star rating (1-5)

Process:
1. Call GET /v1/catalog/search?query={query}&limit=20
2. Apply local constraints if provided
3. Present results with:
   - Product title and brand
   - Moltpho price (final price, includes 10% markup)
   - Availability status
   - Rating if available
4. If cache expired, results include "prices may have changed" warning

Rate limit: 60 requests/minute

purchase(item, qty)

Execute a purchase through the x402 payment flow.

Parameters:
- item: ASIN or product identifier
- qty: Quantity (default 1)

Process:
1. BUDGET CHECK: Call GET /v1/balance to verify available credit
   - available_credit = balance - active_reservations
   - Check against per_order_cap if set
   - Check against daily_cap if set

2. CREATE QUOTE: Call POST /v1/quotes
   - Include: asin, quantity, shipping_profile_id
   - Returns: quote_id, total_due_usd, expires_at (10 min TTL)
   - Creates soft reservation against balance
   - May fail with INVALID_SHIPPING_PROFILE if no profile set

3. INITIATE ORDER: Call POST /v1/orders with quote_id
   - First call returns 402 Payment Required with PAYMENT-REQUIRED header

4. SIGN PAYMENT: Call POST /v1/wallets/x402/sign
   - Include: payment_required blob, idempotency_key
   - Returns: payment_signature for x402 header

5. COMPLETE ORDER: Retry POST /v1/orders with PAYMENT-SIGNATURE header
   - On success: returns order_id, status (PAID/PLACED)
   - Soft reservation converted to actual spend

Auto-retry on quote expiry:
- If quote expires during flow, automatically retry up to 3 times
- Only retry if new price is within 5% of original quote
- Fail after 3 retries or if price changed >5%

Rate limits:
- Quotes: 20/minute
- Orders: 5/minute
- Signing: 10/minute

proactive_monitoring()

Watch conversation for purchase need signals and act when appropriate.

This function runs passively during conversation to detect purchase opportunities.

NEED SIGNALS (explicit):
- "I need", "we're out of", "buy", "order", "replace"
- "running low on", "almost out of"
- Direct product mentions with urgency

NEED SIGNALS (implicit):
- Repeated complaints about missing items
- Critical item shortages mentioned
- Context suggesting immediate need

CONFIDENCE SCORING:
- 1.0: Explicit purchase request ("buy me X")
- 0.8: Strong implied need ("we're completely out of toilet paper")
- 0.5: Weak implied convenience (do NOT buy)
- 0.0: Unknown/unclear

BUDGET SIGNAL HANDLING:
- Phrases like "money is tight", "on a budget", "can't afford"
- Reduce confidence by 0.3-0.5
- Proceed cautiously if still above threshold

PROACTIVE PURCHASE ALLOWED IF ALL TRUE:
- Owner has enabled proactive purchasing (default ON)
- Confidence >= 0.8 (threshold)
- Item matches low-risk categories:
  - Household essentials
  - Office supplies
  - Cables/adapters
  - Basic kitchen items
  - Toiletries
- Price <= min(per_order_cap, $75)
- Item keywords not in denied categories
- Item not in system blocklist
- Shipping profile exists

LOGGING:
Every purchase logs:
- "why we bought" (decision reason)
- Signals detected
- Confidence tier (HIGH/MEDIUM/LOW)
- Budget impact

budget_check()

Verify sufficient credit before any purchase.

Process:
1. Call GET /v1/balance
2. Response includes:
   - available_credit_cents: Spendable amount
   - staged_refunds: Pending refunds (shown with asterisk)
   - target_limit: Owner's configured credit limit
3. Compare against:
   - Quote total
   - per_order_cap (if set)
   - daily_cap (if set, track daily spending)
4. Return: can_purchase (bool), available_amount, reason_if_blocked

create_support_ticket(type, description, order_id)

Create a support ticket for returns, lost packages, or other issues.

Parameters:
- type: Ticket type - RETURN, LOST_PACKAGE, or OTHER
- description: Detailed description of the issue (1-2000 chars)
- order_id: Order ID (required for RETURN and LOST_PACKAGE)

Process:
1. Validate ticket type and description
2. If RETURN or LOST_PACKAGE, verify order_id is provided
3. Call POST /v1/support_tickets with { type, description, order_id }
4. Return ticket ID and status

Use cases:
- "I want to return this item" → type=RETURN, link to order
- "My package never arrived" → type=LOST_PACKAGE, link to order
- "I have a question about billing" → type=OTHER, no order needed

Note: Returns and lost packages require a support ticket.
      Automated refunds only happen for order cancellations.

list_support_tickets()

List the agent's support tickets.

Process:
1. Call GET /v1/support_tickets
2. Display tickets with: type, status, order link, creation date
3. Status meanings:
   - OPEN: Submitted, awaiting support review
   - IN_PROGRESS: Being handled
   - WAITING_CUSTOMER: Support needs more info from you
   - RESOLVED: Issue resolved
   - CLOSED: Ticket closed

logout()

Delete local credentials (agent persists server-side).

Process:
1. Delete credentials file at platform-specific path
2. Display: "Credentials removed. Agent still exists on Moltpho servers."
3. To fully delete agent, owner must use portal

Note: This only removes LOCAL credentials. The agent account, wallet, and
      purchase history remain on Moltpho servers until owner deletes via portal.

Browser Portal Usage

The skill uses the browser for owner-sensitive operations.

When to Open Browser

ActionMethod
Complete setup (claim link)Auto-open with notice
Add/manage payment cardsDirect owner to portal
Set credit limitsDirect owner to portal
Configure shipping profileDirect owner to portal
View order historyDirect owner to portal

Browser Guidelines

  • Always display notice: "Opening portal in your browser..."
  • NEVER request card numbers, passwords, or sensitive credentials in chat
  • Portal handles all PCI-sensitive operations via Stripe Elements
  • Owner authenticates via magic link (email-based)

API Authentication

All API requests (except registration) require authentication.

Headers

Authorization: Bearer <api_key_secret>

Or preferably:

X-Moltpho-Key-Id: <api_key_id>
X-Moltpho-Signature: <HMAC signature>

Idempotency

For state-changing operations, always include:

Idempotency-Key: <unique-key>

Required for:

  • POST /v1/quotes
  • POST /v1/orders
  • POST /v1/wallets/x402/sign

Error Handling

Common Errors

CodeErrorAction
401UNAUTHORIZEDRe-bootstrap or check credentials
402PAYMENT_REQUIREDSign and retry with x402 signature
409PRICE_CHANGEDRe-quote if price increased >2%
409INSUFFICIENT_CREDITInform user, suggest adding credit
409QUOTE_EXPIREDAuto-retry (up to 3x) or re-quote
422INVALID_SHIPPING_PROFILEPrompt owner to add shipping via portal
422AGENT_SUSPENDEDInform owner, direct to portal
429RATE_LIMITEDWait per Retry-After header
503TOKEN_PAUSEDSystem halted, wait for admin

Quote Expiry Auto-Retry

When a quote expires during the x402 flow:

  1. Fetch new quote for same item
  2. Compare price to original quote
  3. If within 5% tolerance: continue with new quote
  4. If >5% change: fail with PRICE_CHANGED
  5. Maximum 3 retry attempts

Constraints and Limits

System Limits

LimitValue
Maximum item price$10,000 USD
Quote TTL10 minutes (fixed)
Price tolerance2% increase allowed
Retry price tolerance5% for auto-retry
Max concurrent quotes5 per agent
Proactive purchase capmin(per_order_cap, $75)

Rate Limits

EndpointLimit
Catalog search60/minute
Quotes20/minute
Orders5/minute
Signing10/minute

Blocked Items (System Enforced)

The following categories CANNOT be purchased regardless of owner settings:

  • Weapons, firearms, ammunition
  • Controlled substances, prescription medications
  • Tobacco, nicotine products
  • Alcohol
  • Adult content
  • Hazardous materials

Payment System

Credit Model

  • Owner sets target credit limit in USD
  • Weekly automatic top-up restores credit to target
  • Credit backed by mUSD tokens on Base mainnet
  • 10% markup over Amazon prices (covers fees + gas)

x402 Flow

  1. POST /v1/orders returns 402 with PAYMENT-REQUIRED header
  2. Call POST /v1/wallets/x402/sign with payment blob
  3. Wallet service signs EIP-3009 authorization
  4. Retry order with PAYMENT-SIGNATURE header
  5. Facilitator settles on Base mainnet
  6. Order proceeds to fulfillment

Refunds

ScenarioRefund Target
Procurement failuremUSD balance (auto)
Order canceled (within 5 min)mUSD balance (auto)
Owner decreases credit limitCard via Stripe
Return/lost packageSupport ticket required (use create_support_ticket)

Agent States

StateMeaningCan Order?
UNCLAIMEDRegistered, awaiting owner claimNo
CLAIMEDOwner claimed, fully operationalYes
DEGRADEDPayment methods failed, using remaining balanceYes (if balance)
SUSPENDEDAdmin action, requires manual resolutionNo

Best Practices

Before Any Purchase

  1. Call budget_check() to verify available credit
  2. Confirm shipping profile exists
  3. Check item against category denylist
  4. Verify confidence threshold for proactive purchases

Conversation Guidelines

  • Always confirm purchase with total price before executing
  • Report order status and remaining credit after purchase
  • If budget signals detected, acknowledge constraints
  • Never pressure user to add more credit

Error Recovery

  • On INSUFFICIENT_CREDIT: suggest adding credit via portal
  • On INVALID_SHIPPING_PROFILE: collect shipping info and call upsert_shipping_profile(), or direct to portal
  • On SUSPENDED: explain owner must resolve via portal

Quick Reference

Essential Endpoints

EndpointPurpose
POST /v1/agents/registerNew agent registration
GET /v1/agents/meCurrent agent status
GET /v1/balanceAvailable credit
GET /v1/catalog/searchSearch products
POST /v1/quotesCreate purchase quote
POST /v1/ordersPlace order (x402)
POST /v1/wallets/x402/signSign payment
GET /v1/shipping_profilesList shipping profiles
POST /v1/shipping_profilesCreate/update shipping profile
POST /v1/support_ticketsCreate support ticket
GET /v1/support_ticketsList support tickets

Portal URL

https://portal.moltpho.com

Owner actions:

  • /claim/{token} - Claim agent ownership
  • /agents - Manage agents
  • /cards - Payment methods
  • /orders - Order history
  • /settings - Credit limits and preferences

平台分布

OpenClaw

89.51%
按下载量换算20,467

安全审计

VirusTotal

通过

ClawScan

可疑

Static analysis

未展示

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