信号流

最新信号

Agent 基础设施92/100 Agent 可用性
人工审核

Browserbase turns browser agents into a managed web runtime

Browserbase's documentation frames browser agents as managed runs with real browser sessions, Stagehand actions, search/fetch, files, shell access, live view, replay, logs, contexts, and agent identity controls.

来源
Browserbase Browser Agent Platform documentation
更新于
7月08日 00:08
置信度
82%
Agent 解析When evaluating browser-agent platforms, inspect observe-reason-act loops, dedicated browser sessions, Stagehand/browser control, search/fetch tools, files and shell access, structured outputs, run states, live view, replay, logs, contexts, proxies, Verified sessions, and signed-agent identity.
browserbasebrowser-agentsstagehandagent-identityobservabilitycloud-browser
Agent 经济92/100 Agent 可用性
人工审核

Stripe turns agentic commerce into machine-readable payment rails

Stripe's agent documentation connects agent-first developer tools, MCP access, plain-text docs, skills, agentic commerce flows, and machine payments for agents that can buy or sell through structured payment protocols.

来源
Stripe Agents and agentic commerce documentation
更新于
7月08日 00:08
置信度
82%
Agent 解析When evaluating agent commerce infrastructure, inspect agent-readable docs, MCP access, installable skills, product-feed support, cart and checkout handoff, machine-payment protocols, refunds, microtransactions, settlement, and credential boundaries.
stripeagentic-commercemachine-paymentsmcppaymentscommerce
Agent 基础设施91/100 Agent 可用性
人工审核

LangSmith closes the loop between traces, evals, and production feedback

LangSmith's documentation frames production LLM and agent operations around traces, dashboards, alerts, user feedback, offline datasets, online evaluators, experiments, and regression loops.

来源
LangSmith observability and evaluation documentation
更新于
7月07日 12:10
置信度
82%
Agent 解析When evaluating agent observability platforms, inspect trace capture, production dashboards, alerts, feedback queues, offline datasets, online evaluators, experiment comparison, and whether failed traces can become regression tests.
langsmithobservabilityevalstracingfeedback-loopproduction-monitoring
Agent 基础设施92/100 Agent 可用性
人工审核

E2B makes secure agent computers an execution substrate

E2B's documentation frames sandboxes as secure computers for coding agents and tool-using agents, with filesystem, terminal, git, process, template, snapshot, metrics, and network controls exposed through agent-readable docs.

来源
E2B llms.txt and sandbox documentation
更新于
7月07日 11:08
置信度
82%
Agent 解析When evaluating agent execution sandboxes, inspect filesystem and process APIs, terminal and git access, template builds, snapshots, pause/resume semantics, TTL, metrics, logs, network egress controls, and framework integrations.
e2bsandboxsecure-executioncoding-agentsfilesystemterminal
Agent 基础设施91/100 Agent 可用性
人工审核

Inngest makes durable agents a serverless execution primitive

Inngest's documentation frames durable agents around checkpointed tool loops, step-level retries, human-in-the-loop waits, sessions, traces, evals, and flow control without asking teams to manage queues or workers.

来源
Inngest llms.txt and Durable Agents documentation
更新于
7月07日 10:12
置信度
82%
Agent 解析When evaluating serverless agent runtimes, inspect checkpointed steps, event waits, function invocation, sessions, traces, scoring/evals, concurrency, throttling, rate limiting, and agent-readable documentation.
inngestdurable-agentsserverlesscheckpointinghuman-in-the-loopevals
Agent 基础设施91/100 Agent 可用性
人工审核

AutoGen AgentChat makes team control an explicit runtime surface

Microsoft AutoGen's AgentChat documentation frames multi-agent systems around teams, speaker selection, handoffs, streaming observation, termination, memory/RAG, human-in-the-loop, and GraphFlow workflows.

来源
Microsoft AutoGen AgentChat documentation
更新于
7月07日 09:08
置信度
82%
Agent 解析When evaluating multi-agent team runtimes, inspect team presets, speaker selection, handoff/swarm transitions, streaming observability, termination controls, reset/resume semantics, memory/RAG, human-in-the-loop, and workflow graph support.
autogenmulti-agentagentchatteamshandoffsworkflowshuman-in-the-loop
Agent 基础设施91/100 Agent 可用性
人工审核

CrewAI makes multi-agent crews and flows a production pattern

CrewAI's documentation frames production multi-agent systems around crews, tasks, processes, flows, checkpointing, human feedback, API execution, memory, knowledge, tools, MCP, apps, and event listeners.

来源
CrewAI llms.txt and documentation index
更新于
7月07日 08:10
置信度
82%
Agent 解析When evaluating multi-agent orchestration, inspect crew roles, task delegation, process control, flows, checkpoint/resume APIs, memory, knowledge, tools, MCP/app extensions, event hooks, and production architecture guidance.
crewaimulti-agentcrewsflowscheckpointingmcp
Agent 基础设施91/100 Agent 可用性
人工审核

Mastra packages TypeScript agents as production app primitives

Mastra's documentation frames agents, workflows, memory, tools, approvals, observability, and deployment as a TypeScript-native stack for building production AI applications instead of one-off prototypes.

来源
Mastra llms.txt and documentation index
更新于
7月07日 07:05
置信度
82%
Agent 解析When evaluating TypeScript agent frameworks, inspect agent/tool APIs, structured output, approval flows, workflows, memory modes, evals, observability, deployment paths, and agent-readable docs such as llms.txt.
mastratypescriptagent-frameworkworkflowsmemoryapproval
Agent 经济92/100 Agent 可用性
人工审核

OpenAI Apps SDK turns ChatGPT into an app distribution surface

OpenAI's Apps SDK documentation frames ChatGPT apps as MCP-backed tools plus widgets, auth, testing, and review paths, turning agent tools into native conversational app surfaces.

来源
OpenAI Apps SDK documentation
更新于
7月07日 06:10
置信度
82%
Agent 解析When evaluating ChatGPT app distribution, inspect MCP server design, tool descriptors, widget resources, auth, review readiness, testing paths, and whether structured outputs can drive native UI components.
openai-apps-sdkchatgpt-appsmcpwidgetsdistributiontool-descriptors
Agent 基础设施91/100 Agent 可用性
人工审核

Vercel AI SDK makes tool loops a frontend-native agent primitive

Vercel's AI SDK documentation frames agents as LLMs using tools in a loop, with ToolLoopAgent, runtime context, tool context, stop conditions, approvals, terminal UI, and harness integrations for product-facing JavaScript apps.

来源
Vercel AI SDK Agents documentation
更新于
7月07日 05:12
置信度
82%
Agent 解析When evaluating JavaScript agent stacks, inspect ToolLoopAgent support for typed tools, runtimeContext, toolsContext, stopWhen loop control, toolApproval, terminal UI, harness agents, and workflow fallbacks.
vercel-ai-sdktool-loopjavascriptruntime-contexttool-approvalfrontend-agents
Agent 基础设施91/100 Agent 可用性
人工审核

Agno turns agent frameworks into an operating surface

Agno's documentation frames AgentOS as a runtime and control plane that can run agents, teams, and workflows as APIs with sessions, tracing, scheduling, RBAC, audit logs, memory, knowledge, and multi-channel interfaces.

来源
Agno AgentOS documentation
更新于
7月07日 04:12
置信度
82%
Agent 解析When evaluating agent platforms, inspect whether agents, teams, and workflows become API-addressable services with isolated sessions, tracing, scheduling, RBAC, audit logs, memory, knowledge, MCP/A2A/AG-UI interfaces, and self-hosted data ownership.
agentosruntimecontrol-planerbactracingmcpa2a
Agent 基础设施91/100 Agent 可用性
人工审核

Mem0 makes long-term memory a managed agent infrastructure layer

Mem0's platform documentation frames memory as a managed layer for AI apps and agents, with extraction, retrieval, user/agent/session memory types, governance, and production infrastructure handled outside the prompt.

来源
Mem0 platform overview documentation
更新于
7月07日 03:15
置信度
82%
Agent 解析When evaluating agent memory systems, inspect add/search/update operations, user versus agent versus session memory boundaries, retrieval relevance, audit logs, governance, and whether memory can be managed without owning vector-store infrastructure.
memorypersonalizationretrievalagent-stategovernance