The fenbs blogAll articles
Page 12 of 48
- Codex Alternatives: Cloud and Terminal Agents ComparedThe coding agents people pick instead of OpenAI Codex, or next to it, sorted by the reason you would switch: another agent that runs in the cloud while you do something else, or a terminal agent on your own machine. What each vendor documents today, and what carries over when you move. · 8 min read
- Claude Code Hooks Examples: Ten Hooks Worth CopyingTen Claude Code hooks you can paste into settings.json today: protect paths, deny risky commands, format after edits, run tests before Claude stops, log commands, notify you, load context at session start and more. Each with its JSON, a short script, and what the exit code does. · 9 min read
- Datadog MCP Server: Logs and Metrics in Your Coding AgentDatadog runs its own remote MCP server, generally available since March 9, 2026. What its tools cover, how to add it to Claude Code, Cursor and VS Code, how sign-in and permissions work, what data leaves Datadog, and how to turn an alert into a task your agent can finish. · 7 min read
- AI Agents for Customer Service: What They Handle and What They Should NotAI agents are good at four support jobs: triage, drafting replies, answering from a knowledge base and summarizing. Where they should hand over to a person, the escalation rules to write before launch, how to measure whether they are any good, what to tell customers, and where the bugs they find should go. · 8 min read
- Zendesk MCP: Letting an Assistant Read and Triage TicketsZendesk has announced an MCP server of its own, and ships an MCP client today. What each one is, how to connect Claude or another assistant to your tickets in the meantime, why OAuth scopes matter more than the server you pick, and how to keep customer text from steering the assistant. · 7 min read
- HubSpot MCP Server: Connecting Claude and ChatGPT to Your CRMHubSpot runs two MCP servers: a remote one at mcp.hubspot.com that reads and writes CRM data as the signed-in user, and a local developer server for building on HubSpot. How to connect Claude, ChatGPT and Claude Code, how scopes and permissions work, what it can and cannot change, and what to keep an eye on with customer data. · 7 min read
- AI Sales Agents: Research, Follow-Ups and CRM HygieneAn AI sales agent is most useful on the work around the conversation: researching accounts, drafting follow-ups, preparing for meetings and cleaning up the CRM. What to hand it, what a person must approve, the US email and calling rules in plain words, how to measure it, and where the follow-ups that need a person should wait. · 6 min read
- Weekly Status Report Template for Teams and ClientsA weekly status report people actually read fits in five lines: done, next, blocked, decisions and risks. A template to copy, an internal email and a client version, what to leave out, and how to draft it from the board in five minutes. · 7 min read
- Test Case Template With ExamplesA test case needs eight fields: ID, title, preconditions, steps, expected result, actual result, status and the requirement it checks. A template to copy, a spreadsheet layout, six worked examples, how to derive cases from acceptance criteria, and what to check when an AI assistant drafts them. · 8 min read
- AI Agents for Marketing: Content, Campaigns and ReportingAI agents for marketing do five jobs well: research, drafts from a brief, repurposing, campaign QA and reporting summaries. What each looks like, what a person must approve, the US rules on endorsements, reviews, AI claims and email in plain words, and where brand-voice rules should live. · 8 min read
- Perplexity MCP Server: Research Inside Your AgentPerplexity runs an official MCP server with four tools: search, ask, research and reason. What each one returns, how to connect it in Claude Code, Cursor and Claude Desktop, how sign-in and billing work, how its citations come back, and what to do about the web pages it reads. · 7 min read
- Brave Search MCP: Web Search for Your AgentBrave maintains its own MCP server for the Brave Search API, with tools for web, local, image, video and news search, a summarizer and LLM context. What each tool does, where the API key goes, how to set it up in Claude Code, Claude Desktop and VS Code, and when Brave, Perplexity or Firecrawl is the better fit. · 6 min read
- Grafana MCP Server: Dashboards and Alerts for an AgentGrafana Labs maintains mcp-grafana, an open source MCP server for dashboards, datasources, Prometheus and Loki queries, alerting, incidents and OnCall, and runs a hosted Cloud MCP server for Grafana Cloud. What the tools cover, how to scope the service account token, how to run it read-only, and how to set it up in Claude Code and Cursor. · 7 min read
- Cloudflare MCP Servers: Using Them and Hosting Your OwnCloudflare runs a set of remote MCP servers for working with your Cloudflare account, from one server for the whole API to product servers for docs, Workers bindings and observability. It also documents how to host your own MCP server on Workers. What each server does, how to connect, what changed in the Agents SDK, and when a remote server is worth it. · 7 min read
- Canva MCP Server: Designs From Claude and ChatGPTCanva runs two MCP servers: the Canva AI Connector, which lets Claude, ChatGPT and other assistants make and edit your designs, and the Canva Dev MCP server for people building Canva apps. How to connect each, what the tools can create and change, how brand kits and plans fit in, and the limits and permissions to know first. · 8 min read
- Stripe MCP Server: What an Agent Can Do With PaymentsStripe runs a hosted MCP server that lets Claude, Cursor, Codex and other agents read and change your Stripe account. How to connect it, OAuth vs agent API keys, why you start in a sandbox, what the tools reach, and which actions must always wait for a person. · 8 min read
- Amp Code: The Coding Agent Spun Out of Sourcegraph, ExplainedAmp began at Sourcegraph and is now its own company. What the Amp coding agent is today: the web app and orbs, the CLI and editor connection, the Dial of modes and the models behind it, threads and how they are shared, AGENTS.md, MCP, subagents, and how it compares with Claude Code. · 8 min read
- How to Assign Tasks to Team Members Without Losing TrackAssigning a task well takes four decisions: one owner, the right person for the skill and the load, a clear picture of done, and a check-in you both agreed. A message template, how to check in without hovering, assigning to AI assistants, and how to reassign when plans change. · 7 min read
- How to Organize Tasks at Work: A System That Survives MondayA task system works when it has one inbox, a way to sort, four places a task can be, and a short weekly review that keeps it honest. How to set that up, what to do with tasks that sit for weeks, and how AI assistants can add and update tasks without making a mess. · 7 min read
- Project Management for Startups: Just Enough ProcessA startup of two to ten people needs four things from project management: one board, priorities set weekly, decisions written down, and bugs in the same place as everything else. What to skip until later, the signs it is time to add process, and how to work with AI coding agents from the first day. · 7 min read
- Team Norms Examples: Twenty Working Rules Worth Writing DownTwenty example team norms for meetings, response times, code review, decisions, focus time, remote work and AI assistant use, each specific enough to tell when it is broken. How a team agrees on its norms, where to record them so every AI assistant follows them too, and when to review them. · 8 min read
- Architecture Decision Records (ADR): Template and ExamplesAn architecture decision record is a short Markdown file in your repository that says what the team decided about the system’s structure, why, and what follows from it. Nygard’s five-part template, the MADR alternative, numbering and superseding, a filled example, and how to make coding agents read them. · 8 min read
- Incident Postmortem Template: Blameless and UsefulA postmortem is written after the incident is over, to find out why it happened and what will stop it happening again. The six sections, a template to copy, a filled example, how to keep it blameless, and how action items become tracked work that actually gets done. · 9 min read
- Bug Life Cycle in Software Testing: States and Who Moves ThemThe bug life cycle is the set of states a defect passes through from the moment someone reports it to the moment it is closed. The standard states, who is allowed to move a bug into each one, a simpler four-lane version, the ways a bug can end, and how reopening should work. · 7 min read
- Regression Testing Checklist: What to Re-Test Before Each ReleaseRegression testing checks that what worked last release still works after this one. How to choose what to re-test, what to automate and what to leave to a person, a checklist to copy, how every fixed bug becomes a permanent test, and how an AI agent can run the suite and file what fails. · 7 min read
- Vercel MCP Server: Deployments and Logs in Your AgentVercel MCP is Vercel’s official remote MCP server at mcp.vercel.com. What its tools cover, how to add it to Claude Code, Cursor and VS Code, how to scope it to one project, why it has no read-only switch, and what to do about that. · 7 min read
- MongoDB MCP Server: Setup and Read-Only ModeMongoDB’s official MCP server gives an AI agent database tools and Atlas tools. How connection strings differ from Atlas API credentials, how read-only mode and disabled tools work, a least-privilege database user, and setup in Claude Code, Cursor and VS Code. · 8 min read
- User Acceptance Testing (UAT): Checklist and Test ScriptUser acceptance testing is the last check before release, run by the people who will use the software, not by QA. Entry and exit criteria, a one-page UAT plan, a test script template with a filled example, how sign-off works, and how failures become bugs someone fixes. · 9 min read
- Meeting Action Items: A Template That Gets Them DoneAn action item needs three things: a verb, one owner, and a way to tell it is done. A template to copy, a filled example, how to rewrite vague actions, where they go after the meeting, and how to follow up so the same items stop coming back. · 8 min read
- BRD vs PRD vs FRD: Which Requirements Document You NeedA BRD says why the business needs something, a PRD says what the product will do for its users, and an FRD says exactly how the system must behave. Who writes and reads each, a side-by-side table, one-page skeletons, and why a small team usually needs only a light PRD. · 8 min read
- Technical Design Document Template for Small TeamsA technical design document says how a change will be built before anyone builds it: the context, goals and non-goals, the proposal, the alternatives, the risks, the rollout, and the questions still open. A template to copy, a filled example, how to review one, and how it becomes tasks and context for coding agents. · 8 min read
- AI for Project Managers: Twelve Jobs It Can Take Off Your PlateTwelve everyday project management jobs an AI assistant can draft for you, from status reports to release notes, with what you still check on each one, what you never hand over, and why the work itself belongs on a shared board. · 7 min read
- AI Risk Assessment Template: Before You Deploy an AgentSix questions to answer before an AI assistant or agent gets access to your tools: what data it sees, what it can do, who approves, how it fails, what is logged, and how you undo it. A copyable template, a filled example for an agent with write access to a task board, and where to record the result. · 7 min read
- Microsoft Copilot for Project Management: Planner, Teams and LoopMicrosoft 365 Copilot can keep a project’s sources in a notebook, reach other systems through connectors, work tasks in Planner through Planner Agent, and recap meetings in Teams. What each piece does for a project manager, where Copilot on its own stops, and how to connect a board through Copilot Studio. · 8 min read
- Gemini for Project Management in Google WorkspaceGemini can keep a project’s sources in a notebook, save your instructions as a Gem, reach tools such as Asana and Salesforce, and work inside Docs, Sheets, Gmail, Meet and Chat. What each piece does for a project manager, where Gemini on its own stops, and how to connect it to a board the whole team shares. · 8 min read
- Gmail and Google Calendar MCP: Email and Scheduling From an AgentThere are three ways to give an AI assistant your Gmail and Google Calendar: Claude’s built-in connectors, Google’s own Gmail and Calendar MCP servers, and ChatGPT’s Google apps. What each can read and change, which OAuth scopes they ask for, how to set them up, and which actions should always wait for you. · 8 min read
- Outlook MCP Server: Email and Calendar in Claude and CopilotTwo official routes put Outlook mail and calendar behind MCP: Claude’s Microsoft 365 connector and Microsoft’s Work IQ Mail and Calendar servers, which Copilot Studio agents and coding agents use. What each can do, the Entra permissions behind them, delegated vs application access, admin consent, setup, and safe uses. · 7 min read
- Task Tracker Template: The Columns That MatterA task tracker is only as good as its columns. The six that every tracker needs, the date column most teams get wrong, the ones to leave out, a spreadsheet template for Excel or Google Sheets, and the point where a board does the job better. · 7 min read
- Getting Things Done (GTD) on a Kanban BoardDavid Allen’s Getting Things Done method has five steps and a weekly review, and all of it fits on a board with four lanes. How each step maps to a lane, how contexts and projects become labels, what the weekly review looks like on a board, and what changes when a team and its AI assistants share it. · 7 min read
- Bug Tracking Tools for Small Teams: What to Look ForMost bug tracking tools can hold a bug. What separates them for a small team is how fast a bug gets in, what happens to it next, who else can see it, and whether an AI assistant can work with it safely. Seven criteria, the kinds of tool on offer, an honest fit list and the tests to run in a trial. · 7 min read
- AI Agent Use Cases: Forty Examples by TeamForty AI agent use cases across nine teams, from engineering and support to finance, HR and the solo founder. Each is one line: the job, what the agent does, and what a person approves before it counts. · 8 min read
- Claude Code Output Styles: Changing How It TalksOutput styles change how Claude Code responds for a whole session: shorter, more explained, hands-on for learning, or not a software engineer at all. The five built-in styles as documented today, how to write your own, and when a style is the wrong tool. · 7 min read
- Factory Droids: What They Are and How Teams Use ThemFactory is an agent-native software development platform, and Droid is its coding agent. What Droid is today: the droid CLI, the Factory App, editor and Slack surfaces, a choice of models from several labs, AGENTS.md, MCP, four autonomy levels, and how it compares with Claude Code. · 7 min read
- Sequential Thinking MCP: What It Does and Whether You Need ItThe Sequential Thinking server is one of the MCP project’s reference servers, and one of the most installed. What its single tool really does, its status today, how it compares with a model’s built-in thinking and with plan mode, and when it earns its tokens. · 7 min read
- Filesystem MCP Server: Giving an Agent Folder Access SafelyThe reference Filesystem MCP server lets an AI assistant read, write, move and search files in folders you name. How allowed directories and Roots work, the tools it exposes, setup in Claude Desktop, how to keep it read-only, and why coding agents such as Claude Code do not need it. · 7 min read
- Kubernetes MCP Server: Letting an Agent Inspect a Cluster SafelyThe main Kubernetes MCP servers today: the open-source containers/kubernetes-mcp-server, Amazon’s EKS server, Google’s GKE server and Microsoft’s AKS-MCP. How to run each read-only, bind a dedicated service account to the view role, pin one kubeconfig context, keep destructive tools off, and set it up in Claude Code. · 8 min read
- Google Sheets and Excel MCP Servers: Spreadsheets From an AgentThe official ways to give an AI assistant your spreadsheets today: Claude’s Google Sheets and Drive connectors, Google’s own Sheets MCP server, ChatGPT’s Google apps, and, for Excel, Claude’s Microsoft 365 connector and Microsoft’s Work IQ servers. What each can read and change, the scopes behind them, and the three risks spreadsheets add: formulas, shared sheets and instructions hidden in cells. · 8 min read
- AI Agents for Recruiting: Screening Without Bias TrapsAI agents for recruiting are safest on the work around a hiring decision: job descriptions, scheduling, candidate questions and application summaries. What each job looks like, why automated screening and ranking carry legal risk under Title VII and the ADA, what a person must approve, and where the hand-offs should wait. · 8 min read