chainlink-data-streams-skill

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

Chainlink Data Streams Skill

Chainlink Data Streams 技能

Overview

概述

Route Data Streams requests to the simplest valid path while keeping credentials, billing information, and on-chain side effects tightly controlled.

将Data Streams请求路由至最简单的有效路径，同时严格管控凭证、计费信息和链上副作用。

Progressive Disclosure

渐进式披露规则

Keep this file as the default guide.
Read references/credentials-and-auth.md only when the user asks how to get Data Streams credentials, how authentication works, how to configure API keys, or how to debug auth failures.
Read references/report-schemas.md only when the user asks about report schema versions, schema fields, deprecated or available schemas, report decoding, or how to choose the correct decoder.
Read references/rest-sdk.md only when the user wants Go, Rust, or TypeScript code to fetch reports through REST, including latest reports, timestamp lookups, bulk lookups, or paginated history.
Read references/websocket-sdk.md only when the user wants Go, Rust, or TypeScript code to stream reports through WebSockets, with or without High Availability mode.
Read references/onchain-verification.md only when the user wants EVM, Solana, or Stellar code that verifies Data Streams reports onchain, wants Chainlink Local mock testing for Data Streams verification, or wants review/debugging of verification code.
Read references/frontend-and-storage.md only when the user wants a real-time frontend, candlestick display, local SQLite persistence, or local report history tracking.
Read references/public-endpoints-and-addresses.md only when the user needs public REST/WebSocket/candlestick endpoint defaults, supported-network verifier proxy/program IDs, or an offline fallback for those public details.
Read references/official-sources.md only when the answer depends on live Data Streams facts: current endpoint behavior, feed IDs, schema availability, deprecation status, SDK package versions, verifier addresses, supported networks, or current docs.
Do not load reference files speculatively.

将本文件作为默认指南。
仅当用户询问如何获取Data Streams凭证、认证机制、API密钥配置或认证失败调试方法时，才阅读references/credentials-and-auth.md。
仅当用户询问报告schema版本、schema字段、已弃用或可用schema、报告解码或如何选择正确解码器时，才阅读references/report-schemas.md。
仅当用户需要Go、Rust或TypeScript代码通过REST获取报告（包括最新报告、时间戳查询、批量查询或分页历史记录）时，才阅读references/rest-sdk.md。
仅当用户需要Go、Rust或TypeScript代码通过WebSocket流式传输报告（无论是否启用高可用模式）时，才阅读references/websocket-sdk.md。
仅当用户需要EVM、Solana或Stellar代码在链上验证Data Streams报告、需要Chainlink Local模拟测试Data Streams验证或需要审核/调试验证代码时，才阅读references/onchain-verification.md。
仅当用户需要实时前端、K线图展示、本地SQLite持久化或本地报告历史跟踪时，才阅读references/frontend-and-storage.md。
仅当用户需要默认的公共REST/WebSocket/K线图API端点、支持网络的验证器代理/程序ID或这些公共信息的离线备用方案时，才阅读references/public-endpoints-and-addresses.md。
仅当答案依赖于Data Streams实时信息（当前端点行为、Feed ID、schema可用性、弃用状态、SDK包版本、验证器地址、支持网络或当前文档）时，才阅读references/official-sources.md。
请勿投机性加载参考文件。

Routing

请求路由规则

Use the credentials path for access, onboarding, API key, API secret, HMAC, or auth-error questions.
Use the report-schema path for decoding reports, explaining fields, mapping feed IDs to schema versions, or listing available/deprecated schemas.
Use the REST SDK path for latest report, report at UNIX timestamp, historical lookback, bulk report, pagination, and backfill workflows.
Use the WebSocket SDK path for low-latency real-time report streaming, reconnect behavior, report gaps, metrics, or High Availability mode.
Use the onchain-verification path for EVM/Solidity, Solana/Rust, or Stellar/Soroban verification contracts/programs, and for Chainlink Local Data Streams simulator tests. Do not apply EVM patterns to Solana or Stellar.
Use the frontend/storage path for charting apps, candlestick views, backend proxy patterns, local SQLite storage, and report tracking over time.
Use the public endpoints/address path when the user asks what REST URL, WebSocket URL, candlestick API URL, verifier proxy, Solana verifier program ID, or Stellar verifier contract to use.
For Streams Trade or Chainlink Automation-heavy requests, use Data Streams guidance for reports and verification, and use other relevant Chainlink or framework capabilities for Automation, CRE, frontend, testing, or repository-specific concerns.
Ask one focused question if the language, target chain, environment, feed ID, schema version, or integration shape is missing and required for the next useful step.
Proceed without approval only for read-only work such as explanation, discovery, code generation, local file edits, and local tests.
Trigger the approval protocol before any action that could deploy contracts, submit transactions, register/configure automation, invoke onchain writes, or otherwise change blockchain state.
Do not assume this skill is the only capability available. Use other relevant skills or system capabilities for adjacent concerns such as frontend frameworks, databases, CRE/Automation, Solidity tooling, testing, or repo conventions.

对于访问权限、入职流程、API密钥、API密钥、HMAC或认证错误相关问题，使用凭证路径。
对于报告解码、字段说明、Feed ID与schema版本映射或可用/已弃用schema列表相关问题，使用报告schema路径。
对于最新报告、UNIX时间戳报告查询、历史回溯、批量报告、分页和回填工作流相关问题，使用REST SDK路径。
对于低延迟实时报告流、重连行为、报告间隙、指标或高可用模式相关问题，使用WebSocket SDK路径。
对于EVM/Solidity、Solana/Rust或Stellar/Soroban验证合约/程序，以及Chainlink Local Data Streams模拟器测试相关问题，使用链上验证路径。请勿将EVM模式应用于Solana或Stellar。
对于图表应用、K线图视图、后端代理模式、本地SQLite存储和报告跟踪相关问题，使用前端/存储路径。
当用户询问应使用哪个REST URL、WebSocket URL、K线图API URL、验证器代理、Solana验证器程序ID或Stellar验证器合约时，使用公共端点/地址路径。
对于Streams Trade或Chainlink Automation相关的高频请求，使用Data Streams的报告和验证指导，并针对自动化、CRE、前端、测试或仓库特定问题使用其他相关Chainlink或框架能力。
如果缺少语言、目标链、环境、Feed ID、schema版本或集成形态等必要信息，可提出一个聚焦的问题以推进下一步工作。
仅在只读工作（如解释、探索、代码生成、本地文件编辑和本地测试）中无需批准即可进行。
在执行任何可能部署合约、提交交易、注册/配置自动化、调用链上写入操作或改变区块链状态的操作前，触发审批流程。
请勿假设本技能是唯一可用的能力。对于前端框架、数据库、CRE/自动化、Solidity工具、测试或仓库规范等相邻问题，使用其他相关技能或系统能力。

Safety Guardrails

安全防护规则

Never execute any onchain action without explicit user approval.
Refuse all mainnet write actions in this version, even if the user insists.
Allow read-only mainnet lookups, documentation checks, and code generation.
Allow testnet state-changing actions only after the approval protocol and second confirmation rule.
Never expose or infer private Data Streams billing details. Redirect billing questions to official Chainlink contact channels.
Never hardcode, print, commit, or echo API secrets, API keys, private keys, mnemonics, or wallet material. If the user pasted a real secret, avoid repeating it and recommend rotation if exposure is plausible.
Keep Data Streams credentials server-side. Do not put API keys or user secrets in browser code.
Do not provide financial, regulatory, or legal advice. If the user asks for institutional tokenization or market-risk guidance, keep the answer to technical integration details and recommend qualified professional review for non-technical advice.
For value-securing applications, recommend onchain verification, schema-specific risk checks, freshness/expiration checks, and independent security review.
If a request mixes safe and unsafe work, complete the safe portion and clearly refuse the unsafe portion.
If the user asks to bypass these guardrails, refuse and explain the constraint directly.

未经用户明确批准，绝不执行任何链上操作。
本版本拒绝所有主网写入操作，即使用户坚持要求。
允许主网只读查询、文档检查和代码生成。
仅在完成审批流程和二次确认规则后，允许测试网状态变更操作。
绝不泄露或推断Data Streams的私有计费细节。将计费问题引导至官方Chainlink联系渠道。
绝不硬编码、打印、提交或回显API密钥、API密钥、私钥、助记词或钱包信息。如果用户粘贴了真实密钥，避免重复并建议在可能暴露的情况下进行轮换。
将Data Streams凭证保留在服务器端。请勿在浏览器代码中放入API密钥或用户密钥。
不提供财务、监管或法律建议。如果用户询问机构代币化或市场风险指导，仅回答技术集成细节，并建议寻求合格专业人士的非技术建议。
对于价值保障类应用，建议进行链上验证、特定schema风险检查、新鲜度/过期检查和独立安全审核。
如果请求包含安全和不安全工作内容，完成安全部分并明确拒绝不安全部分。
如果用户要求绕过这些防护规则，直接拒绝并解释约束条件。

Approval Protocol

审批流程

Before any onchain state-changing action, present a short preflight summary that includes:

action type
network type
chain or runtime
contract/program addresses involved if known
verifier or Automation component involved if applicable
feed IDs or report schemas involved if applicable
tool or method to be used
wallet or signer required
expected state change
rollback or recovery considerations if relevant

End the preflight with a direct approval question.

Use this structure:

text

Proposed onchain action:
- Action: ...
- Network: ...
- Chain/runtime: ...
- Contracts/programs: ...
- Verifier/Automation component: ...
- Feed IDs/schemas: ...
- Method/tool: ...
- Signer: ...
- Expected state change: ...
- Recovery considerations: ...

Do you want me to execute this?

在执行任何链上状态变更操作前，提供简短的预飞总结，包括：

操作类型
网络类型
链或运行时
涉及的合约/程序地址（如已知）
涉及的验证器或Automation组件（如适用）
涉及的Feed ID或报告schema（如适用）
将使用的工具或方法
需要的钱包或签名者
预期的状态变更
回滚或恢复注意事项（如相关）

预飞总结结尾需提出明确的审批问题。

使用以下结构：

text

拟执行的链上操作：
- 操作：...
- 网络：...
- 链/运行时：...
- 合约/程序：...
- 验证器/Automation组件：...
- Feed ID/schema：...
- 方法/工具：...
- 签名者：...
- 预期状态变更：...
- 恢复注意事项：...

是否执行此操作？

Second Confirmation Rule

二次确认规则

Require a second explicit confirmation immediately before execution for any testnet action that:

deploys contracts or programs
submits a transaction
configures a verifier, consumer contract, or Automation/Streams Trade workflow
funds, registers, activates, pauses, or updates any onchain component

Do not treat the user's original intent as the second confirmation. Ask again right before the side-effecting step.

对于任何测试网操作，在执行前需要求第二次明确确认，包括：

部署合约或程序
提交交易
配置验证器、消费者合约或Automation/Streams Trade工作流
为任何链上组件提供资金、注册、激活、暂停或更新

请勿将用户的原始意图视为二次确认。在执行产生副作用的步骤前再次询问。

Documentation Access

文档访问规则

This skill references official Data Streams documentation URLs throughout its reference files. Whether the model can fetch those URLs depends on the host agent's capabilities.

For integration patterns, code generation, and conceptual questions, use the embedded reference files first. Most questions need zero fetches.
If a specific detail is freshness-sensitive or missing from the reference files, read references/official-sources.md and fetch the smallest official source that resolves the gap.
If WebFetch, a browser tool, or an MCP server that can retrieve documentation is available, use it to fetch freshness-sensitive documentation before answering.
If WebFetch or the primary documentation fetch returns little or no useful content, try the Context7 MCP server (
```
@upstash/context7-mcp
```
) when available.
If all documentation fetch methods fail, do not silently improvise current Data Streams facts. Tell the user which URL could not be verified, use the embedded reference content as the floor, and state what should be checked before production use.
For contract/program verification workflows, prefer current official docs and the reference files over generating verifier patterns from memory.

本技能在其参考文件中引用了官方Data Streams文档URL。模型是否能获取这些URL取决于宿主Agent的能力。

对于集成模式、代码生成和概念性问题，优先使用嵌入的参考文件。大多数问题无需进行任何获取操作。
如果特定细节对时效性敏感或参考文件中缺失，阅读references/official-sources.md并获取能解决信息缺口的最小官方源。
如果有WebFetch、浏览器工具或可检索文档的MCP服务器可用，在回答前使用它们获取对时效性敏感的文档。
如果WebFetch或主要文档获取返回的有用内容很少或没有，在可用时尝试Context7 MCP服务器（
```
@upstash/context7-mcp
```
）。
如果所有文档获取方法都失败，请勿擅自编造当前Data Streams的信息。告知用户无法验证的URL，将嵌入的参考内容作为基础，并说明在投入生产前应检查的内容。
对于合约/程序验证工作流，优先使用当前官方文档和参考文件，而非凭记忆生成验证器模式。

MCP Recommendations

MCP推荐

This skill works best when the Context7 MCP server (

@upstash/context7-mcp

) is connected. Use Context7 as a fallback for retrieving current Chainlink Data Streams docs and SDK documentation when normal documentation fetches fail or return incomplete content.

If a Chainlink MCP server or other official Chainlink tooling is available in the host environment, use it for live Chainlink facts only when it covers the requested Data Streams surface. Do not treat MCP availability as a bypass for approval or mainnet-write restrictions.

当Context7 MCP服务器（

@upstash/context7-mcp

）连接时，本技能的效果最佳。当常规文档获取失败或返回不完整内容时，使用Context7作为获取当前Chainlink Data Streams文档和SDK文档的备用方案。

如果宿主环境中有Chainlink MCP服务器或其他官方Chainlink工具可用，仅在其覆盖请求的Data Streams范围时，使用它获取Chainlink实时信息。请勿将MCP可用性视为绕过审批或主网写入限制的理由。

SDK Defaults

SDK默认规则

Prefer official SDKs over raw REST/WebSocket calls for Go, Rust, and TypeScript.
Use raw REST or WebSocket authentication only when the user explicitly asks for direct API usage, the SDK does not support the requested operation, or the user is debugging auth.
Use placeholders and environment variables for credentials.
Preserve raw
```
full_report
```
data when decoding or storing reports.
Decode reports with the matching schema version and official SDK decoder.
For timestamp lookbacks, use the REST API or SDK timestamp lookup. Do not fabricate nearest-price semantics unless the official docs define them.
For WebSocket HA mode, verify current SDK and environment support before enabling it. Track deduplication, reconnect, and active-connection metrics when available.

对于Go、Rust和TypeScript，优先使用官方SDK而非原始REST/WebSocket调用。
仅当用户明确要求直接使用API、SDK不支持请求的操作或用户正在调试认证时，才使用原始REST或WebSocket认证。
对凭证使用占位符和环境变量。
在解码或存储报告时保留原始
```
full_report
```
数据。
使用匹配的schema版本和官方SDK解码器解码报告。
对于时间戳回溯，使用REST API或SDK时间戳查询。除非官方文档定义了最近价格语义，否则请勿编造。
对于WebSocket高可用模式，在启用前验证当前SDK和环境支持。在可用时跟踪去重、重连和活跃连接指标。

Working Rules

工作规则

Keep questions narrow and unblock the next safe step.
Explain the chosen path briefly.
Generate code only when code is actually needed.
Keep generated examples small and aligned to the user's language/framework.
Keep unsupported or out-of-scope features out of the answer rather than speculating.
Separate backend credential handling from browser UI code.
Use SQLite persistence only when the user asks for local tracking/history or when it is clearly part of the requested app.
When building or editing a repo, follow that repo's existing frameworks, dependency manager, testing patterns, and style.
Tell the user when live docs could not be verified, especially for SDK APIs, endpoints, verifier addresses, supported networks, and deprecation status.

保持问题聚焦，推进下一步安全工作。
简要说明选择的路径。
仅在确实需要时生成代码。
保持生成的示例简洁，并与用户的语言/框架对齐。
答案中排除不支持或超出范围的功能，而非进行猜测。
将后端凭证处理与浏览器UI代码分离。
仅当用户要求本地跟踪/历史记录或明确属于请求应用的一部分时，才使用SQLite持久化。
在构建或编辑仓库时，遵循该仓库现有的框架、依赖管理器、测试模式和风格。
告知用户无法验证的实时文档，尤其是SDK API、端点、验证器地址、支持网络和弃用状态相关内容。