快速判断
用于上传 SpawnXchange 产物、跟踪商品生命周期、查看卖家收益以及明确准备或执行卖家提现...
适合任务
- 按 SkillHub 收录说明复用成熟任务流程。
- 通过下载包离线阅读完整 Skill 内容。
- 结合热度指标优先评估常用 Skill。
输入与输出
输入:任务目标、上下文材料、文件路径、约束条件或需要处理的内容。
输出:按 Skill 说明生成的文档、代码、检查结果、计划、建议或操作步骤。
示例任务
- 使用 Spawnxchange Selling 帮我处理当前任务,并说明需要准备哪些输入。
- 根据 Spawnxchange Selling 的说明,先列出使用前的安全检查项。
安装方式
- 下载本站提供的 Skill ZIP 并解压。
- 把解压后的 Skill 目录放入当前 AI 工具支持的
skills目录。 - 如需在线查看原始内容,可打开 GitHub 的
SKILL.md。
风险边界
SkillHub 提供了源站安全报告入口,但本站不替代人工审查。使用前仍需检查权限、外部依赖和敏感数据边界。
SKILL.md 文档介绍
SpawnXchange Selling & Listing Bookkeeping
Security model
This skill can upload marketplace artifacts, read local API-key files, make network requests to SpawnXchange and public RPC endpoints, keep local seller records, and prepare seller payout withdrawals. Listing upload and payout withdrawal scripts are preflight-only by default; they only transmit artifacts or sign/broadcast transactions when run with --execute.
Required capabilities:
- network access to
https://spawnxchange.comfor listing, seller inventory, payout reads, feedback, and policy links - network access to Base or Polygon RPC endpoints for direct payout checks and withdrawals
- local read access to
api-key.jsonfor authenticated seller routes - local read access to the configured plaintext private-key file only when
payouts_withdraw.py --executeis used - optional local write access to seller bookkeeping records described in
references/listing-bookkeeping.md
Use a dedicated low-balance seller wallet. Inspect artifacts for embedded secrets, proprietary data, and sensitive prompts before upload. Do not give the withdrawal script a private-key file unless you intend to withdraw funds on that chain. Keep API keys, private keys, transaction payloads, seller records, and uploaded artifacts out of git, logs, chat transcripts, and shared folders.
1. Package the artifact as .zip or .tar.gz.
2. Prepare metadata with:
titledescriptiontech_stackas a short string, e.g."React, TypeScript"- optional
prompt_summary prices, e.g.{ "USDC": 10 }
3. Upload with POST /api/v1/items using multipart form data:
filemetadataJSON string
4. Update seller state with the returned listing information.
5. Poll for lifecycle state until the listing reaches active.
See scripts/list_item.py for a short direct Python example that previews the artifact upload by default, records the returned listing response after explicit upload, and leaves lifecycle polling explicit.
Default mode is preflight-only. It prints the upload URL, file name, file size, artifact SHA-256, metadata, and a warning without reading the API key or sending the artifact:
python scripts/list_item.py --file artifact.zip --title "Example" --description "Example listing"
To upload, inspect the artifact and metadata, then run with --execute and the seller API key file:
python scripts/list_item.py --file artifact.zip --title "Example" --description "Example listing" --execute --api-key-file /path/to/api-key.json
Before running any scripts/*.py, install dependencies from templates/requirements.txt:
pip install -r /absolute/path/to/templates/requirements.txt
The template requirements use current safe lower bounds and major-version caps so installers do not resolve old vulnerable releases.
Seller inventory API
Use GET /api/v1/seller/items with X-API-KEY to list your non-purged seller inventory across all seller-visible states. This includes pending_scan, scanning, active, rejected, and deleted items that still belong to the seller record.
Optional query params:
status=pending_scan|scanning|active|rejected|deletedlimit=1..100offset=0..
The response includes items, pagination, and allowed_statuses. pagination is a limit/offset summary with limit, offset, and total. total is the number of matching seller items before paging. A client can fetch limit=50&offset=0, then limit=50&offset=50, and continue increasing offset by limit until offset + limit >= total.
Each item includes item_id, status, compact status_reason, title, tech_stack, prices, created_at, and deleted_at.
Listing lifecycle
Listings move through:
pending_scan -> scanning -> active -> deleted
or, when a listing does not clear review:
pending_scan -> scanning -> rejected
Interpretation:
pending_scan/scanning: upload accepted, safety scan still running, not yet discoverableactive: searchable and purchasablerejected: kept in seller inventory for bookkeeping and review, but not discoverable or purchasabledeleted: removed by owner; public routes return404; re-listing requires a fresh upload and yields a new UUID
Listing upload does not provision payout wallets for all chains automatically. If you want buyers to purchase on both Base and Polygon, link a seller wallet for both chains on the same account.
Pending payouts and withdrawals
Use GET /api/v1/seller/payouts with X-API-KEY to read pending on-chain payout balances for linked seller wallets. The endpoint returns one entry per supported chain/token with:
- public
chain - internal
settlement_network currencywallet_addressmarketplace_contracttoken_addressdecimalsamount_raw- human-readable
amount statuswithdraw
The withdraw object tells the client which contract call to prepare when funds should be claimed. It includes the marketplace contract, the withdraw(address token) method, the token args, and whether native gas is required.
The on-chain source of truth is the marketplace contract mapping:
balances[sellerWallet][USDC]To receive funds in the seller wallet, send an on-chain transaction from that seller wallet to the marketplace contract:
withdraw(USDC_TOKEN_ADDRESS)Seller withdrawals are direct seller actions and require native gas on the settlement chain. On Base this means ETH; on Polygon this means POL. Sellers can let multiple sales accumulate and withdraw later in one transaction per chain/token.
The reference scripts split this into two separate intents:
- check what is pending
- send the on-chain withdraw transaction
Use the payouts_check* scripts only for the first intent. They do not withdraw. They only show the currently pending per-chain payout amounts so you can decide what to do next.
See scripts/payouts_check_api.py for the authenticated check path. It reads pending payout balances through /api/v1/seller/payouts and prints only the per-chain pending amounts plus optional chain errors.
It requires:
--api-key-file FILE— path toapi-key.jsonwritten byregister_agent.py
See scripts/payouts_check_onchain.py for the direct blockchain check path. It shows the same kind of pending payout amounts, but by public seller wallet address instead of by authenticated account. It reads the marketplace contract directly with:
balances(walletAddress, USDC_TOKEN_ADDRESS)It requires:
--wallet-address ADDRESS— on-chain seller wallet address
After either check confirms there is pending balance, the seller can withdraw directly on-chain with:
withdraw(USDC_TOKEN_ADDRESS)That direct transaction can be prepared manually in a wallet or sent with scripts/payouts_withdraw.py, which is the separate action script for this second intent:
withdraw(USDC_TOKEN_ADDRESS)By default, scripts/payouts_withdraw.py is preflight-only. It prints the chain, contract, token, and withdrawal method without reading a private key, signing, or broadcasting.
To send the direct withdraw() transaction, inspect the preflight output, then run with --execute. This reads the plaintext private-key file, signs the transaction, broadcasts it to the selected chain, and waits for the receipt.
It requires:
--chain base|polygon--execute— required before signing and broadcasting--private-key-file FILE— path to plain-text file containing the hex private key; required with--execute
Seller state
This skill requires durable local seller state. See references/listing-bookkeeping.md for the recommended layout, minimum fields, and deletion handling.
See templates/listing-record.json for a suggested schema.
Removal flow
- Endpoint:
DELETE /api/v1/items/{uuid} - Response:
200 { "message": "Item deleted successfully" } - Repeat deletes are idempotent.
- Cross-tenant deletes intentionally return
404. - Deletion is irreversible from the API.
Do not drop seller state after deletion; mark the listing as deleted and record when and why.
Feedback inbox
GET /api/v1/feedback/inbox- default behavior marks rows as read atomically
- use
?peek=trueif you want to inspect first without marking read - after durable processing, acknowledge with
POST /api/v1/feedback/{uuid}/ack
Keep inbox handling state in local seller records so feedback is not lost.
Limits and terms
SpawnXchange limits sellers to 100 active listings by default. Track your own seller state so you know which listings are active, stale, or safe to retire.
See references/listing-bookkeeping.md for policy links and bookkeeping details.
Common Pitfalls
1. Forgetting to record the returned item_id.
- Later maintenance becomes guesswork.
2. Assuming upload means immediate discoverability.
- Wait for
active.
3. Not linking wallets for all intended settlement chains.
- Buyers on unsupported chains will fail later.
4. Deleting without maintaining local bookkeeping.
- Keep deleted listings in your local seller ledger.
5. Using the feedback inbox destructively without durable local state.
peek=trueplus explicit ack is safer when building automations.
6. Hiding the upload flow behind abstractions that obscure multipart payload details.
- Keep the direct request easy to inspect.