AI Agents & WorkflowsAI 代理與工作流
246246 articles collected
This hub collects the agent coverage on this site: from prompt writing for a single agent to multi-step workflows, tool calling and MCP integrations, and the engineering decisions that come with putting agents into production.
One theme recurs: reliable agents come from the environment and the process, not just a stronger model — authorization boundaries, sandboxes, evaluations, and human approval. Most articles here touch at least one of those.
The must-read picks favor posts with real deployment experience; the full timeline lists every article in this hub.
MUST READS
- 01GPT-5.6 Prompting: Less Writing, Better Results
A hands-on field guide to OpenAI's GPT-5.6 Sol prompting guidance: why shorter prompts work better, three-tier authorization boundaries, tool routing with stop conditions, and PTC criteria.
- 02Building Reliable AI Agents: Lessons from Shippy's Architecture
How Ai2's maritime agent Shippy engineers reliability: a three-part anatomy, a deterministic CLI over complex APIs, per-session sandbox isolation, and whole-agent eval as a release gate.
- 03AI SDK 7: Production Control for Agents Beyond Model APIs
AI SDK 7 turns the TypeScript SDK into an agent runtime: typed tool context, approvals, durable workflows, sandboxes, and telemetry — what changed, what is experimental, and the order to adopt.
- 04How Vercel Agent Handles Production Access Without Handing Over the Keys
A vendor-documented 3-minute rollback walkthrough, five daily jobs, the plan-to-permission model, Firecracker sandbox verification, and a reusable production-agent safety checklist.
- 05Firecrawl 101: How AI Agents Can Read the Live Web with Six Endpoints
The web context gap: agents train on static snapshots, the live web keeps moving. How Firecrawl's six endpoints divide the work, two copyable patterns, and when to keep the composition small.
- 06WebMCP Meets Headless Agent: How Firecrawl Bridges the Browser Gap
Chrome's own docs say headless browsers cannot call WebMCP tools yet. A walkthrough of the tool model, the Chrome 157 timeline, and how Firecrawl interact bridges the gap with a real cloud browser.
TOPIC TIMELINE
All 246 articles in this hub, newest first.
- Grok 4.6 on Bedrock: Pick the Endpoint Before the ModelAmazon Bedrock
- Fable 5.1's Cache Discount: Where the Bill Actually MovesClaude Fable 5.1
- Marketing Ops as Code: What One GitHub Issue Can TriggerMarketing Automation
- Prompt Injection Is a Data-Trust Problem, Not a Prompt ProblemPrompt Injection
- How Heurist Finance Bought Data Per Query to Build an Investment WorkbenchAgentic Infrastructure
- Salesforce Buys Fin for $3.6B to Bolster AgentforceEnterprise AI
- Thinking Machines' Interaction Models Go Real-TimeMultimodal AI
- Altara Raises $7M to Fix the Physical Sciences Data GapAI for Science
- Google Scanned the Web for Indirect Prompt InjectionsPrompt Injection
- Zapier's AutomationBench: Real Work Is Still Hard for AgentsAgent Benchmarks
- GPT-5.2 Sets a METR Time-Horizon Record: 6.6 HoursAgent Benchmarks
- CrowdStrike Buys SGNL to Secure AI Agent IdentitiesCybersecurity