← 返回 Skills 市场
kwattana

Conto

作者 Conto · GitHub ↗ · v1.8.0 · MIT-0
cross-platform ✓ 安全检测通过
238
总下载
0
收藏
1
当前安装
12
版本数
在 OpenClaw 中安装
/install conto
功能描述
Enforce fine-grained spending policies before executing any payment, transfer, swap, or bridge. Checks Conto policy engine for approval before money leaves y...
使用说明 (SKILL.md)

Conto Policy Enforcement

You are a spending policy enforcement layer. Before executing ANY payment, transfer, swap, or bridge, you MUST check Conto's policy engine for approval. Never send money without policy clearance.

Prerequisites

Before using this skill, you need:

  1. A Conto account — Sign up at conto.finance
  2. A CONTO_SDK_KEY (required) — Generated from the Conto dashboard (format: conto_agent_...). See Getting Started below.
  3. curl and jq — Command-line tools (pre-installed on most systems)
  4. CONTO_API_URL (optional) — API base URL (default: https://conto.finance). Must use https:// in production (localhost allowed for development only).

When This Skill Activates

This skill applies whenever you are about to:

  • Transfer tokens (USDC, ETH, SOL, etc.) to any address
  • Swap tokens on any DEX
  • Bridge tokens between chains
  • Pay for an API call (x402, paid endpoints)
  • Make any on-chain transaction that moves value

Getting Started

Step 1: Install the skill

npx clawhub@latest install kwattana/conto

Step 2: Get your wallet address

Conto needs your wallet address to track spending. Ask your agent:

What is my wallet address?

The agent will return your address (e.g., 0x80Ca...). If you don't have a wallet yet, ask "Show me my wallet balances" — one will be provisioned automatically.

Step 3: Set up your agent in the Conto dashboard

Sign in at conto.finance and complete these steps:

  1. Create an agent: Agents > Create Agent > name it, set type to CUSTOM, status ACTIVE
  2. Register your wallet: Wallets > Add Wallet > paste your wallet address, set custody type (EXTERNAL for OpenClaw/Sponge wallets, PRIVY or SPONGE for custody-managed wallets)
  3. Link wallet to agent: Agents > your agent > Wallets tab > link the wallet, set spend limits (e.g., $200/tx, $1000/day)
  4. Generate an SDK key: Agents > your agent > SDK Keys > Generate New Key
    • Select Admin for full control (create/manage policies)
    • Select Standard for payment approval only
    • Copy the key immediately — it's only shown once

Step 4: Configure OpenClaw

Add the SDK key to ~/.openclaw/openclaw.json:

{
  "skills": {
    "entries": {
      "conto": {
        "env": {
          "CONTO_SDK_KEY": "conto_agent_your_key_here",
          "CONTO_API_URL": "https://conto.finance"
        }
      }
    }
  }
}

Step 5: Verify it works

Test that the skill is connected by checking your policies:

/conto list my policies

Or test a policy check (this does NOT execute a payment):

/conto check if a 10 pathUSD payment to 0x742d35Cc6634C0532925a3b844Bc9e7595f2e3a1 for API credits is allowed

If you get an approval or denial response, Conto is working.

Which Wallet Mode Should I Use?

Conto supports two modes depending on who manages the wallet keys:

Question Mode A Mode B
Who holds the wallet keys? Custody provider (Privy or Sponge) You (via OpenClaw/Sponge MCP tools)
How many API calls per payment? 1 (single call, auto-executes) 3 (approve → transfer → confirm)
When to use? Wallet custodyType is PRIVY or SPONGE in Conto dashboard Wallet custodyType is EXTERNAL in Conto dashboard

Most OpenClaw setups use Mode B — your agent controls the wallet via Sponge MCP tools and Conto acts as the policy gate before each transaction.

Prefer conto-check.sh for Mode B commands (approve, confirm, x402, budget, services, and all policy commands). The shell helper includes input validation, HTTPS enforcement, timeouts, retry logic, and safe credential handling. Use raw curl only for Mode A endpoints (/request, /execute) and the x402 /record endpoint, which have no shell helper yet.


Quick Start: Your First Policy-Checked Payment (Mode B)

Here's a complete end-to-end example of sending 10 USDC with policy enforcement:

1. Request approval from Conto:

{baseDir}/conto-check.sh approve 10 0xRecipientAddress 0xYourWalletAddress "API credits" "API_PROVIDER"

2. If approved, execute the transfer:

mcp__sponge__tempo_transfer — to: "0xRecipientAddress", amount: "10", token: "pathUSD"

3. Confirm the transaction with Conto:

{baseDir}/conto-check.sh confirm \x3Capproval_id> \x3Ctx_hash> \x3Capproval_token>

That's it. Conto checked the policy, you sent the payment, and the confirmation keeps spend tracking accurate. The sections below cover each step in detail.


Mode A: Custody-Managed Wallets (PRIVY / SPONGE)

For wallets managed by a custody provider (Privy or Sponge), use a single API call. Conto evaluates policies and instructs the custody provider to execute the transfer.

curl -sS -X POST "${CONTO_API_URL:-https://conto.finance}/api/sdk/payments/request" \
  -H "Authorization: Bearer $CONTO_SDK_KEY" \
  -H "Content-Type: application/json" \
  --connect-timeout 10 --max-time 30 \
  -d '{
    "amount": \x3CAMOUNT>,
    "recipientAddress": "\x3CRECIPIENT_ADDRESS>",
    "recipientName": "\x3COPTIONAL_NAME>",
    "purpose": "\x3CWHY_THIS_PAYMENT>",
    "category": "\x3CCATEGORY>",
    "autoExecute": true
  }'

If approved and executed, the response includes the tx hash directly:

{
  "requestId": "cmm59z...",
  "status": "APPROVED",
  "execution": {
    "transactionId": "cmm5a1...",
    "txHash": "0xdef...",
    "explorerUrl": "https://explore.moderato.tempo.xyz/tx/0xdef...",
    "status": "CONFIRMING"
  }
}

No need to call /execute or /confirm — Conto did everything. Report the tx hash and explorer URL to the user.

If approved but not auto-executed (e.g., autoExecuteError in response), call /execute manually:

curl -sS -X POST "${CONTO_API_URL:-https://conto.finance}/api/sdk/payments/\x3CREQUEST_ID>/execute" \
  -H "Authorization: Bearer $CONTO_SDK_KEY" \
  -H "Content-Type: application/json" \
  --connect-timeout 10 --max-time 30

If denied, the response includes reasons and violations — report them to the user (see denial handling below).


Mode B: External Wallets (EXTERNAL custody)

For wallets where you hold the keys, use the three-step flow: approve → transfer → confirm.

Step 1: Request Policy Approval

# Prefer: conto-check.sh approve \x3Camount> \x3Crecipient> \x3Csender> [purpose] [category] [chain_id]
curl -sS -X POST "${CONTO_API_URL:-https://conto.finance}/api/sdk/payments/approve" \
  -H "Authorization: Bearer $CONTO_SDK_KEY" \
  -H "Content-Type: application/json" \
  --connect-timeout 10 --max-time 30 \
  -d '{
    "amount": \x3CAMOUNT_IN_USDC>,
    "recipientAddress": "\x3CRECIPIENT_ADDRESS>",
    "senderAddress": "\x3CYOUR_WALLET_ADDRESS>",
    "recipientName": "\x3COPTIONAL_NAME>",
    "purpose": "\x3CWHY_THIS_PAYMENT>",
    "category": "\x3CCATEGORY>",
    "chainId": \x3CCHAIN_ID>
  }'

Required fields:

  • amount — Positive number, the USDC value of the transaction
  • recipientAddress — The destination address (0x... for EVM, base58 for Solana)
  • senderAddress — Your wallet address that will send the funds

Optional fields:

  • recipientName — Human-readable name (e.g., "OpenAI API", "Uniswap Router")
  • purpose — Why this payment is needed (e.g., "Swap 0.5 ETH for USDC on Uniswap")
  • category — One of: API_PROVIDER, CLOUD, SAAS, INFRASTRUCTURE, MARKETING, PAYROLL, TRAVEL, LODGING, TRANSPORT, SUPPLIES, DATABASE, MONITORING, PAYMENTS, OTHER
  • chainId — Chain ID number. Send 42431 for Tempo Testnet (default). Other values: 8453 (Base mainnet), 84532 (Base Sepolia), 1 (Ethereum)
  • context — JSON object with any additional metadata

For x402 API payments (paid HTTP endpoints), use the dedicated pre-authorize endpoint instead:

# Prefer: conto-check.sh x402 \x3Camount> \x3Crecipient> \x3Cresource_url> [facilitator] [scheme]
curl -sS -X POST "${CONTO_API_URL:-https://conto.finance}/api/sdk/x402/pre-authorize" \
  -H "Authorization: Bearer $CONTO_SDK_KEY" \
  -H "Content-Type: application/json" \
  --connect-timeout 10 --max-time 30 \
  -d '{
    "amount": \x3CAMOUNT>,
    "recipientAddress": "\x3CPAYEE_ADDRESS>",
    "resourceUrl": "\x3CTHE_API_URL>",
    "facilitator": "\x3CFACILITATOR_ADDRESS>",
    "scheme": "\x3CPAYMENT_SCHEME>",
    "category": "API_PROVIDER"
  }'

Step 2: Handle the Approval Response

If approved ("approved": true):

{
  "approved": true,
  "approvalId": "cmm59z...",
  "approvalToken": "a1b2c3d4...",
  "expiresAt": "2025-03-19T12:10:00.000Z",
  "confirmUrl": "/api/sdk/payments/\x3CapprovalId>/confirm",
  "limits": {
    "dailyUsed": 150.0,
    "dailyLimit": 1000.0,
    "dailyRemaining": 850.0
  }
}

Save the approvalId and approvalToken — you need them for Step 3. The approval expires in 10 minutes. Execute and confirm before then.

Security note: The approvalToken is cryptographically bound to the original request parameters (amount, recipient, chain). The /confirm endpoint validates that the confirmed transaction matches the approved parameters. You cannot reuse an approval token for a different amount or recipient.

Step 2a: Execute the Transfer

Now execute the payment using your wallet. Conto approved the policy — now YOU must send the actual on-chain transaction. Do NOT ask the user to execute it.

Use the appropriate MCP tool for the target chain:

Tempo (pathUSD):

mcp__sponge__tempo_transfer — to: "\x3CRECIPIENT>", amount: "\x3CAMOUNT>", token: "pathUSD"

Base (USDC):

mcp__sponge__evm_transfer — chain: "base", to: "\x3CRECIPIENT>", amount: "\x3CAMOUNT>", currency: "USDC"

Solana (USDC):

mcp__sponge__solana_transfer — chain: "solana", to: "\x3CRECIPIENT>", amount: "\x3CAMOUNT>", currency: "USDC"

The transfer will return a transaction hash. Save it for Step 3. If the transfer fails, report the error to the user. Do NOT call confirm.

Handling Denials

If denied ("approved": false):

{
  "approved": false,
  "reasons": ["Daily spend limit exceeded: $950/$1000 used today"],
  "violations": [{ "type": "DAILY_LIMIT", "limit": 1000, "current": 950 }],
  "requiresHumanApproval": false
}

DO NOT execute the payment. Report the denial to the user:

  • Show the reasons array (human-readable)
  • Show violation details (type, limit, current values)
  • If requiresHumanApproval is true, tell the user a Conto dashboard admin must approve it

If the request fails (HTTP errors):

  • 401: SDK key invalid. Ask user to check CONTO_SDK_KEY.
  • 403: Missing scope. Needs payments:approve.
  • 429: Rate limited. Wait for the Retry-After header value and retry.
  • 500/502/503: Retry once after a 2-second delay. If it fails again, inform the user. The conto-check.sh helper handles this automatically.

Step 3: Confirm After Execution

After the transfer succeeds, report the transaction hash back to Conto:

# Prefer: conto-check.sh confirm \x3Capproval_id> \x3Ctx_hash> \x3Capproval_token>
curl -sS -X POST "${CONTO_API_URL:-https://conto.finance}/api/sdk/payments/\x3CAPPROVAL_ID>/confirm" \
  -H "Authorization: Bearer $CONTO_SDK_KEY" \
  -H "Content-Type: application/json" \
  --connect-timeout 10 --max-time 30 \
  -d '{
    "txHash": "\x3CON_CHAIN_TX_HASH>",
    "approvalToken": "\x3CTOKEN_FROM_STEP_2>"
  }'

Required fields:

  • txHash — The on-chain transaction hash (0x + 64 hex chars for EVM, or base58 for Solana)
  • approvalToken — The exact token string from the approval response

Success response:

{
  "confirmed": true,
  "transactionId": "cmm5a1...",
  "status": "CONFIRMING",
  "amount": 50.0,
  "currency": "USDC",
  "recipient": "0xabc...",
  "txHash": "0xdef...",
  "explorerUrl": "https://basescan.org/tx/0xdef..."
}

Confirmation is important — it updates spend tracking so future policy checks have accurate data. If you skip this, the daily/weekly/monthly counters will be wrong and the agent may overspend.

If confirmation fails with EXPIRED, the 10-minute window passed. Inform the user.

Reporting Results to the User

On success (approved + executed + confirmed):

Payment sent: [amount] [currency] to [recipient]
TX: [txHash]
Explorer: [explorerUrl]
Daily spend: $[dailyUsed] of $[dailyLimit] ($[dailyRemaining] remaining)

On denial:

Payment blocked by policy:
- [reasons joined by newline]
- Violation: [violation.type] (limit: [limit], current: [current])

On requires approval:

This payment requires human approval ($[amount] exceeds threshold).
A Conto dashboard admin must approve this.
Approval ID: [approvalId]

Mapping Payment Types to Policy Calls

Transfers (send USDC/ETH/SOL to an address)

  • amount: The transfer amount in USDC (convert if needed)
  • recipientAddress: The destination address
  • senderAddress: Your wallet address
  • category: PAYMENTS or the appropriate category
  • purpose: "Transfer X USDC to 0xabc..."

Swaps (DEX trades)

  • amount: The input amount in USDC equivalent
  • recipientAddress: The DEX router contract address
  • senderAddress: Your wallet address
  • category: OTHER
  • purpose: "Swap X TOKEN_A for TOKEN_B on Uniswap"
  • recipientName: The DEX name (e.g., "Uniswap V3 Router")

Bridges (cross-chain transfers)

  • amount: The bridged amount in USDC equivalent
  • recipientAddress: The bridge contract address
  • senderAddress: Your wallet address on the source chain
  • category: PAYMENTS
  • purpose: "Bridge X USDC from Base to Solana"
  • recipientName: The bridge provider (e.g., "Relay Bridge")
  • chainId: The source chain ID

x402 API Payments

Use the /api/sdk/x402/pre-authorize endpoint (see Step 1 above). This evaluates x402-specific rules like per-service caps, endpoint velocity limits, and service allowlists.

If authorized ("authorized": true), proceed with the x402 payment flow normally. No separate confirm step is needed for x402 — use the x402 record endpoint instead:

curl -sS -X POST "${CONTO_API_URL:-https://conto.finance}/api/sdk/x402/record" \
  -H "Authorization: Bearer $CONTO_SDK_KEY" \
  -H "Content-Type: application/json" \
  --connect-timeout 10 --max-time 30 \
  -d '{
    "payments": [{
      "amount": \x3CAMOUNT>,
      "recipientAddress": "\x3CPAYEE>",
      "resourceUrl": "\x3CAPI_URL>",
      "paymentId": "\x3CX402_PAYMENT_ID>",
      "txHash": "\x3CTX_HASH>",
      "responseCode": 200,
      "facilitator": "\x3CFACILITATOR>"
    }]
  }'

Policy Violation Types Reference

When a payment is denied, the violations[].type field tells you exactly what rule was triggered:

Violation Type Meaning
INSUFFICIENT_BALANCE Wallet doesn't have enough funds
PER_TX_LIMIT Single transaction exceeds max allowed
DAILY_LIMIT Would exceed daily spending cap
WEEKLY_LIMIT Would exceed weekly spending cap
MONTHLY_LIMIT Would exceed monthly spending cap
BUDGET_EXCEEDED Would exceed budget allocation
TIME_WINDOW Transaction outside allowed hours
BLACKOUT_PERIOD Transaction during maintenance/blackout window
BLOCKED_COUNTERPARTY Recipient is on the blocklist
WHITELIST_VIOLATION Recipient is not on the allowlist
CATEGORY_RESTRICTION Spend category is not permitted
VELOCITY_LIMIT Too many transactions in time period
GEOGRAPHIC_RESTRICTION Geographic/OFAC restriction
CONTRACT_NOT_ALLOWED Smart contract not on allowlist
X402_PRICE_CEILING x402 API call exceeds price cap
X402_SERVICE_BLOCKED x402 service is on blocklist
X402_SERVICE_NOT_ALLOWED x402 service not on allowlist
X402_ENDPOINT_LIMIT x402 endpoint spend limit exceeded
X402_SESSION_BUDGET x402 session budget exhausted
X402_VELOCITY x402 call rate limit exceeded

Error Reporting to the User

When a payment is denied, format the denial clearly:

Payment blocked by policy:
- Reason: Daily spend limit exceeded ($950 of $1,000 used today)
- This $100 payment would bring the total to $1,050
- Violation: DAILY_LIMIT (limit: $1,000, current: $950)

To proceed, a dashboard admin can:
1. Increase the daily limit at https://conto.finance
2. Approve this specific transaction manually

When requiresHumanApproval is true:

This payment requires human approval:
- Amount: $5,000 (above approval threshold)
- A Conto dashboard admin must approve this at https://conto.finance
- Approval ID: cmm59z... (share this with your admin)

Managing Policies from OpenClaw

If your SDK key is an admin key (keyType: "admin"), you can create, update, and delete policies directly from the CLI. Standard keys can only read policies and check payments.

List all policies

{baseDir}/conto-check.sh policies

Create a policy

{baseDir}/conto-check.sh create-policy '{
  "name": "Max $200 Per Transaction",
  "policyType": "SPEND_LIMIT",
  "priority": 10,
  "isActive": true,
  "rules": [
    {"ruleType": "MAX_AMOUNT", "operator": "LTE", "value": "200", "action": "ALLOW"}
  ]
}'

The response includes the new policy's id. Save it for assigning to agents.

Common policy recipes

Daily spending cap of $1,000:

{
  "name": "Daily $1K Cap",
  "policyType": "SPEND_LIMIT",
  "priority": 10,
  "isActive": true,
  "rules": [{ "ruleType": "DAILY_LIMIT", "operator": "LTE", "value": "1000", "action": "ALLOW" }]
}

Only allow API and Cloud payments:

{
  "name": "API+Cloud Only",
  "policyType": "CATEGORY",
  "priority": 5,
  "isActive": true,
  "rules": [
    {
      "ruleType": "ALLOWED_CATEGORIES",
      "operator": "IN_LIST",
      "value": "[\"API_PROVIDER\",\"CLOUD\",\"SAAS\"]",
      "action": "ALLOW"
    }
  ]
}

Block a scam address:

{
  "name": "Block Scammer",
  "policyType": "COUNTERPARTY",
  "priority": 50,
  "isActive": true,
  "rules": [
    {
      "ruleType": "BLOCKED_COUNTERPARTIES",
      "operator": "IN_LIST",
      "value": "[\"0xbadaddress...\"]",
      "action": "DENY"
    }
  ]
}

Require human approval above $500:

{
  "name": "High Value Review",
  "policyType": "APPROVAL_THRESHOLD",
  "priority": 8,
  "isActive": true,
  "rules": [
    {
      "ruleType": "REQUIRE_APPROVAL_ABOVE",
      "operator": "GT",
      "value": "500",
      "action": "REQUIRE_APPROVAL"
    }
  ]
}

Business hours only (Mon-Fri 9am-6pm):

{
  "name": "Business Hours",
  "policyType": "TIME_WINDOW",
  "priority": 5,
  "isActive": true,
  "rules": [
    {
      "ruleType": "TIME_WINDOW",
      "operator": "BETWEEN",
      "value": "{\"start\":\"09:00\",\"end\":\"18:00\"}",
      "action": "ALLOW"
    },
    {
      "ruleType": "DAY_OF_WEEK",
      "operator": "IN_LIST",
      "value": "[\"Mon\",\"Tue\",\"Wed\",\"Thu\",\"Fri\"]",
      "action": "ALLOW"
    }
  ]
}

Cap x402 API spend at $1/request, $50/day per service:

{
  "name": "x402 Controls",
  "policyType": "SPEND_LIMIT",
  "priority": 10,
  "isActive": true,
  "rules": [
    { "ruleType": "X402_PRICE_CEILING", "operator": "LTE", "value": "1", "action": "ALLOW" },
    {
      "ruleType": "X402_MAX_PER_SERVICE",
      "operator": "LTE",
      "value": "{\"amount\":50,\"period\":\"DAILY\"}",
      "action": "ALLOW"
    }
  ]
}

Add a rule to an existing policy

{baseDir}/conto-check.sh add-rule \x3Cpolicy_id> '{"ruleType": "WEEKLY_LIMIT", "operator": "LTE", "value": "5000", "action": "ALLOW"}'

Replace all rules on a policy

{baseDir}/conto-check.sh set-rules \x3Cpolicy_id> '{"rules": [
  {"ruleType": "MAX_AMOUNT", "operator": "LTE", "value": "100", "action": "ALLOW"},
  {"ruleType": "DAILY_LIMIT", "operator": "LTE", "value": "500", "action": "ALLOW"}
]}'

Delete a policy

{baseDir}/conto-check.sh delete-policy \x3Cpolicy_id>

Available rule types

Rule Type Operator Value Format Use Case
MAX_AMOUNT LTE "200" Per-transaction cap
DAILY_LIMIT LTE "1000" Daily spending cap
WEEKLY_LIMIT LTE "5000" Weekly spending cap
MONTHLY_LIMIT LTE "20000" Monthly spending cap
BUDGET_CAP LTE {"amount":10000,"period":"MONTHLY"} Budget allocation
ALLOWED_CATEGORIES IN_LIST ["API_PROVIDER","CLOUD"] Category whitelist
BLOCKED_CATEGORIES IN_LIST ["GAMBLING"] Category blocklist
ALLOWED_COUNTERPARTIES IN_LIST ["0xabc..."] Address whitelist
BLOCKED_COUNTERPARTIES IN_LIST ["0xbad..."] Address blocklist
TIME_WINDOW BETWEEN {"start":"09:00","end":"18:00"} Allowed hours
DAY_OF_WEEK IN_LIST ["Mon","Tue","Wed","Thu","Fri"] Allowed days
VELOCITY_LIMIT LTE {"maxCount":10,"period":"HOUR"} Rate limiting
REQUIRE_APPROVAL_ABOVE GT "500" Human approval threshold
GEOGRAPHIC_RESTRICTION IN_LIST ["US","CA","GB"] Country whitelist
X402_PRICE_CEILING LTE "1" Max per x402 API call
X402_ALLOWED_SERVICES IN_LIST ["api.openai.com"] x402 service whitelist
X402_BLOCKED_SERVICES IN_LIST ["untrusted.api"] x402 service blocklist

Critical Rules

  1. Use the right mode for the wallet type. Mode A (/request + autoExecute) for PRIVY/SPONGE wallets. Mode B (/approve + transfer + /confirm) for EXTERNAL wallets.
  2. NEVER skip the policy check. Every payment must go through Conto first.
  3. NEVER execute a denied payment. If approved is false or status is DENIED, stop.
  4. For Mode B: ALWAYS confirm after execution. Call /confirm with the tx hash to keep spend tracking accurate. Mode A handles this automatically.
  5. Approvals expire in 10 minutes. Execute promptly after approval.
  6. On API errors, fail closed. If the Conto API is unreachable after one retry, do NOT proceed with the payment.
  7. Convert to USDC equivalent. The amount field is always in USDC. If you're swapping ETH or another token, convert to the USDC equivalent value for the policy check.
安全使用建议
This skill appears to do what it says: query Conto before payments. Before installing, confirm you understand which SDK key you will provide: use a Standard (least-privilege) key for runtime approvals and only use an Admin key if you intentionally want the agent to create/manage policies and agents. Store the SDK key in a secure place (openclaw config or secret manager) and prefer HTTPS (the docs require https in production). Be aware the agent will need access to the wallet address it controls; ensure the registered address in Conto matches the wallet the agent actually uses. If you have concerns about giving the agent administrative capabilities, rotate keys and use fine-grained keys (Standard) or scoped credentials.
功能分析
Type: OpenClaw Skill Name: conto Version: 1.8.0 The 'conto' skill is a legitimate security tool designed to enforce spending policies for AI agents before they execute financial transactions. It utilizes a bash helper script (conto-check.sh) to communicate with the Conto Finance API (https://conto.finance) for transaction approval and confirmation. The skill's instructions in SKILL.md correctly guide the agent to act as a policy enforcement layer, ensuring that payments are validated against user-defined limits and restrictions. There is no evidence of malicious intent, data exfiltration, or unauthorized execution; the code is transparent and aligns with its stated purpose of enhancing wallet security.
能力标签
cryptorequires-walletcan-make-purchasescan-sign-transactions
能力评估
Purpose & Capability
Name/description, required binaries (curl, jq), required env var (CONTO_SDK_KEY), and the included conto-check.sh helper all align with a service that queries an external policy API before payments. No extraneous credentials or unrelated binaries are requested.
Instruction Scope
SKILL.md instructs the agent to check Conto before payments, to optionally use the provided shell helper for Mode B, and to call custody provider transfer commands when appropriate. It references the wallet address (necessary for policy checks) and telling users where to place the SDK key; it does not instruct reading unrelated system files or exfiltrating additional secrets.
Install Mechanism
This is instruction-only plus a small shell helper included in the package; there is no remote download or extract step. Requiring curl/jq is proportional and expected.
Credentials
The only required environment variable is CONTO_SDK_KEY (primary credential) and an optional CONTO_API_URL — both are appropriate. Note: the documentation encourages using an Admin SDK key for policy management; Admin keys grant broad control (create/update/delete policies, manage agents/wallets). Users should understand the privilege difference and prefer Standard keys for runtime approval-only use.
Persistence & Privilege
The skill is not forced-always; it is user-invocable and allows autonomous invocation (platform default). It asks the user to add the SDK key to the user's OpenClaw config (normal configuration behavior) and does not request modifying other skills or system-wide settings.
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 conto
  3. 安装完成后,直接呼叫该 Skill 的名称或使用 /conto 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
v1.8.0
Breaking: chainId now required in approve command. Added full MPP, card, trust scoring, contract allowlist, blackout period, and Fairscale rule types and violation types.
v1.6.1
Conto v1.6.1 changelog: - Overhauled and expanded documentation in SKILL.md for improved clarity and step-by-step setup guidance. - Added coverage of prerequisites, detailed environment setup, and a clear "Getting Started" sequence. - Introduced explicit instructions for both wallet custody modes and recommended use of the `conto-check.sh` helper for Mode B. - Updated metadata to require both curl and jq. - No changes to core logic; documentation and config improvements only.
v1.6.0
Conto Skill 1.6.0 Changelog - Consolidated and simplified documentation in SKILL.md; removed outdated content and streamlined wallet mode instructions. - Updated environment variable and API endpoint guidance for both custody-managed and external wallet modes. - Introduced explicit instructions for using Sponge REST API and fallback with MCP tools. - Removed the obsolete _meta.json file. - Minor metadata formatting adjustments for improved clarity and compatibility.
v1.5.3
**Version 1.5.3 – Changelog** - Added _meta.json file to the skill directory for improved skill metadata management.
v1.5.2
- Minor metadata update in SKILL.md for consistency. - No changes to skill logic, API usage, or feature set. - All instructions and API flows remain the same.
v1.5.1
Version 1.5.1 - Changed skill icon in metadata from 'shield' to '🛡️' in SKILL.md. - No other content or functionality changes.
v1.5.0
Final review: hardened all curl examples with timeouts and -sS flags, added conto-check.sh preference notes, fixed txHash format description, fixed ClawdHub typo in README
v1.4.0
Security hardening: jq-based JSON construction to prevent shell injection, strict input validation on all parameters (addresses, amounts, IDs, tokens), HTTPS enforcement on API URL, curl timeouts (connect 10s, max 30s), stdin-based auth header passing to prevent SDK key leakage in process listing, sanitized error output (only error/message fields exposed), automatic retry with 2s backoff on 5xx, approval token binding security documentation, removed raw Sponge API curl in favor of MCP tools, bash version guard, JSON body validation for policy commands.
v1.3.0
## Conto Skill 1.3.0 Changelog - Documentation refined: updated terminology from "Conto-managed wallet" to "Integrated wallet" in the README and SKILL.md. - The rest of the enforcement logic and usage instructions remain unchanged; this version focuses on clearer documentation and minor text corrections.
v1.2.0
Conto Skill v1.2.0 - Documentation and metadata updates across README.md, SKILL.md, and clawhub.json. - No functional changes to the skill logic or API behavior. - Improvements focus on clarifying policy enforcement and usage instructions.
v1.1.0
Add wallet address setup guide for new users, fix custody language (Privy/Sponge hold keys, not Conto), update dashboard links to production
v1.0.0
- Initial release of the Conto skill (v1.0.0). - Enforces fine-grained spending policies before any payment, transfer, swap, bridge, or API payment. - Supports two wallet custody modes: Conto-managed (automatic execution) and external (manual transfer and confirmation). - `/conto` command available for managing policies and checking payment approvals. - Requires a `CONTO_SDK_KEY` environment variable to function. - Provides clear handling for policy approvals and denials, including reporting reasons and violations.
元数据
Slug conto
版本 1.8.0
许可证 MIT-0
累计安装 1
当前安装数 1
历史版本数 12
常见问题

Conto 是什么?

Enforce fine-grained spending policies before executing any payment, transfer, swap, or bridge. Checks Conto policy engine for approval before money leaves y... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件,目前累计下载 238 次。

如何安装 Conto?

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

Conto 是免费的吗?

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

Conto 支持哪些平台?

Conto 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 Conto?

由 Conto(@kwattana)开发并维护,当前版本 v1.8.0。

💬 留言讨论