3 min read

AgentKey Product Update: July–October 2026

AgentKey Product Update: July–October 2026

Researching a company, finding papers, checking job listings, or planning a trip all require information beyond what a model already knows.

From July through October, we expanded AgentKey’s client support and data coverage, introduced a unified Scenario API in Beta, and improved tool discovery and billing visibility. These updates give agents more capabilities to work with through a single AgentKey account.

More Ways to Connect Your Agent

AgentKey now offers a native Codex plugin, a Kimi Code plugin, a Gemini CLI extension, and plugin support for Cursor and Antigravity. We’ve also added a dedicated MCP integration for DeepSeek Harness.

Our supported client list now includes Grok, Grok Bot, and Grok Build, giving users more ways to access external data through the agents they already use.

For plugins and extensions that support native authorization, setup uses a browser sign-in flow. We’ve also fixed Claude Code’s native authorization experience and improved Gemini CLI’s installation package and first-time connection process.

These changes reduce the setup work between installing AgentKey and making the first data call.

See supported clients and connection guides →

New Data for More Everyday Tasks

The catalog now covers more sources for academic research, job searches, travel planning, and weather queries.

  • Academic research: alphaXiv and AMiner expand access to papers, publication details, researchers, and institutions for literature reviews and technical research.
  • Jobs and recruitment: Liepin and JSearch support job searches and listing lookups, helping agents collect and organize recruitment information.
  • Travel planning: Booking data services and Rollinggo add hotel, flight, and destination information for researching and comparing travel options.
  • Weather: QWeather and Google Weather provide weather information for travel preparation and local queries.

An agent researching a trip can use these sources to gather accommodation options, flight information, and weather conditions. These are information-retrieval capabilities; they do not imply automatic booking or purchasing.

Broader Business, Financial, and Social Coverage

We’ve added RootData for Web3 project, funding, organization, and people data. SEC / SEC-API, World Bank, and Eurostat expand access to company disclosures and economic information.

Social and content coverage has grown too. New providers including Rnote, Sorsa, and RedFox broaden access to Xiaohongshu, X, WeChat content, and cross-platform trends. Additional integrations extend coverage to Telegram and automotive content platforms, including Dongchedi, Yiche, and Autohome.

An agent researching a company can draw on business records, market information, and public conversations. An agent preparing a content brief can gather material across platforms instead of relying on a single feed.

The available tasks and fields depend on the source.

A Unified Scenario API for Developers

Developers building AI applications and workflows can now access AgentKey’s data capabilities through the Scenario API Beta.

As of October 26, the production catalog lists 184 Beta methods across 10 categories: social, e-commerce, finance, crypto, business, web, travel, jobs, weather, and academia.

Comparable tasks use standardized inputs and result formats, with pricing defined at the method level. Teams can integrate around the task they need to perform while reducing the work of maintaining separate provider interfaces.

For methods configured with multiple available sources, the service can attempt an alternative when a source fails. The method price stays the same. Responses also identify the data source, the sources attempted, and the actual credits consumed.

The OpenAPI documentation provides method definitions, request parameters, and response structures to help developers get started.

Explore the Scenario API Beta →

Easier Discovery and Clearer Usage Costs

We’ve improved natural-language tool discovery, including recognition of Chinese platform names and aliases, so agents can more easily find capabilities relevant to a request.

Users can also control capabilities by category. When AgentKey cannot provide a result, the updated Skill supports the agent continuing with other tools available in its client.

For usage-based tools, we’ve added clearer cost explanations. Execution results report the actual credits charged, helping users and agents understand what each call costs.

To try these updates, connect AgentKey to your preferred client, or explore the Scenario API for your own application.

The Scenario API is currently in Beta. Supported tasks and fields vary by method and source. Data queries do not imply automatic booking, purchasing, or job-application submission.