# AgenticKit Blog

> Guides on Cursor rules, slash commands, AI agent architecture, and shipping SaaS with AI coding workflows.

Canonical: https://www.agentickit.co/blog
Last updated: 2026-11-18

- [Best Claude Code MCP Servers for Solo SaaS Founders in 2026 (Ranked)](https://www.agentickit.co/blog/best-mcp-servers-claude-code-2026-ranked) - GitHub, Postgres/Supabase, Stripe, Sentry, and Context7 rank as the core five; Playwright and Chrome DevTools earn a spot only once you're shipping UI. A ranked top-7 with a decision table, not a catalog.
- [Best MCP Servers for Claude Code in 2026 (Curated, Not Context-Hungry)](https://www.agentickit.co/blog/best-mcp-servers-claude-code-not-context-hungry-2026) - GitHub, Postgres/Supabase, and Context7 earn a permanent seat in context; Stripe and Sentry earn one conditionally. A full PM suite, Slack, and an everything-CRM usually don't. The context-cost filter, applied as a skip list.
- [Best Claude Code MCP Servers for Solo SaaS Founders: 2026 Stack (Stripe, Sentry, Postgres, GitHub)](https://www.agentickit.co/blog/best-claude-code-mcp-servers-saas-founders-starter-stack) - GitHub, Postgres/Supabase, Stripe, Sentry, and Context7 — the five MCP servers worth connecting before any others, with a week-1 vs. month-2 install order and the context-budget case against connecting more.
- [What Is a Cursor Skill? The Plain-English Guide (Plus When You Actually Need One)](https://www.agentickit.co/blog/what-is-a-cursor-skill) - A Cursor skill is a SKILL.md file the agent loads only when your task matches its description — not an always-on rule, not a typed command. Here's what that means in practice, with a decision guide for when you actually need one.
- [MCP Server Rot: How to Vet Any 'Best MCP Servers for Claude Code 2026' List](https://www.agentickit.co/blog/mcp-server-rot-best-mcp-servers-lists-2026) - 'Best MCP servers' roundups get written once and read for months while the servers they recommend keep changing underneath them. Here's the 3-check audit — archived status, last commit date, maintainer org — to run before you wire any of them into Claude Code.
- [Best Claude Code Skills for Solo Founder Marketing & SEO in 2026](https://www.agentickit.co/blog/best-claude-code-skills-marketing-seo-2026) - Dev-focused 'best Claude Code skills' lists only cover frontend-design and webapp-testing. Here are the 11 skills for positioning, on-page SEO, content clusters, and growth loops that a solo founder running their own GTM actually needs.
- [Best MCP Servers for a Solo SaaS Founder's Claude Code Stack (2026)](https://www.agentickit.co/blog/best-mcp-servers-solo-saas-stack-2026) - Supabase/Postgres, Firecrawl, Sentry, Vercel, Stripe, Parallel Search, and the Anthropic reference memory server — picked by job for a one-person SaaS stack, with the per-request cost tradeoffs that decide which ones to leave connected.
- [Best MCP Servers for Claude Code SaaS Founders in 2026](https://www.agentickit.co/blog/best-mcp-servers-claude-code-saas-founders) - Stripe, Supabase/Postgres, GitHub, Playwright, and Sentry MCP servers ranked for a solo founder's Claude Code stack, with the exact claude mcp add command and how to audit a third-party server before connecting it.
- [Best Cursor MCP Servers for Solo SaaS Founders in 2026](https://www.agentickit.co/blog/best-cursor-mcp-servers-saas-founders) - Stripe, Supabase/Postgres, GitHub, and Playwright MCP servers ranked for a solo founder's Cursor stack, with the exact .cursor/mcp.json entry for each and which 'best MCP' lists are written by the vendor.
- [Best Cursor Background Agents for Solo Founders in 2026 (Ranked)](https://www.agentickit.co/blog/best-cursor-background-agents-solo-founders) - PR drafting, long refactors, test-fixing, doc generation, and dependency upgrades ranked for how a one-person SaaS team actually hands off work to Cursor's background agents, plus the cost tradeoffs of running several at once.
- [Cursor Rules vs Skills vs Commands vs Hooks: The Only Diagram You Need (2026)](https://www.agentickit.co/blog/rules-vs-skills-vs-commands-vs-hooks) - Who decides it runs, and how often: rules always, commands when you type them, skills when the agent matches a description, hooks when an event fires.
- [Best Claude Code Commands for Shipping a SaaS MVP in 2026](https://www.agentickit.co/blog/best-claude-code-commands-examples) - A Claude Code command is a markdown file in .claude/commands/ with optional frontmatter for argument-hint, allowed-tools, and model. 12 commands, a full /ship example, and when a command should be a skill instead.
- [13 Best Free Cursor Tools in 2026: Rules Generators, Token Counters & Config Converters](https://www.agentickit.co/blog/free-cursor-tools-2026) - Most "cursor rules generator" results are template libraries, not generators. 13 free tools ranked, what each one actually outputs, and a 30-second pick-one flowchart.
- [Best Cursor Commands in 2026: 12 Copy-Paste Slash Commands for Solo Founders](https://www.agentickit.co/blog/best-cursor-commands) - A Cursor command is a .md file in .cursor/commands. 12 copy-paste examples for solo founders, where they live, and how to fix commands that don't appear.
- [Best Cursor Slash Commands in 2026: 20 Commands Every Solo Founder Should Steal](https://www.agentickit.co/blog/best-cursor-slash-commands) - Slash commands vs rules vs skills in one paragraph, 20 commands grouped by build/verify/operate/grow, and 3 most founders forget to install.
- [25 Cursor Slash Commands and Rules Combos Power Users Swear By in 2026](https://www.agentickit.co/blog/cursor-slash-commands-rules-combos) - Why pairing a .cursor/commands file with a scoped .cursor/rules/*.mdc file beats either alone, with 25 copy-paste combos for billing, auth, testing and CI.
- [Best Cursor Rules for Next.js and SaaS in 2026 (.mdc Examples You Can Copy)](https://www.agentickit.co/blog/best-cursor-rules-2026) - Ten glob-scoped .mdc rules for the solo-SaaS stack, why .cursorrules is legacy, and which rules belong in skills instead.
- [Cursor Rules Generator Prompts: 15 Copy-Paste Examples for 2026](https://www.agentickit.co/blog/cursor-rules-generator-prompts) - 15 ready-made .mdc rule snippets across Next.js, Stripe, Postgres and auth, the mistakes that make rules useless, and a 30-second generator instead of hand-typing them.
- [RIPER-5 in Cursor: The Research-to-Review Workflow, Setup Guide and Rule File](https://www.agentickit.co/blog/riper-5-cursor-workflow) - RIPER-5 stops agents from jumping straight to code. Copy-paste .mdc rule and skill, how it fits Plan Mode and a memory bank, model routing by phase, and a Stripe billing walkthrough.
- [Cursor Playwright Skill: A SKILL.md for Sign-up, Checkout and Dashboard E2E Tests](https://www.agentickit.co/blog/cursor-playwright-skill) - A complete SKILL.md for Playwright in Cursor: selector order, shared fixtures, Stripe boundaries and a flake checklist, plus install path and an example run.
- [Best Cursor Skills in 2026: What SKILL.md Is and 12 Worth Installing](https://www.agentickit.co/blog/best-cursor-skills) - A Cursor skill is a SKILL.md the agent loads when relevant. Rules vs skills vs commands, your first skill in two minutes, and 12 copy-paste skills for solo founders.
- [25 Best Cursor Skills for Solo SaaS Founders in 2026 (Ranked)](https://www.agentickit.co/blog/25-best-cursor-skills-2026-ranked) - 25 skills from AgenticKit's catalog, ranked for solo founders by how often you'd hit the trigger case and how expensive it is if the skill isn't there — grouped by shipping, SEO, infra, and billing.
- [Cursor Skills vs Claude Code Skills: What Actually Transfers Between Them (2026)](https://www.agentickit.co/blog/cursor-skills-vs-claude-code-skills) - A Cursor skill and a Claude Code skill share the exact same SKILL.md format — what breaks when you move one between tools is the folder path and relative references, not the content.
- [Best Cursor Skills for Stripe Billing and Subscription SaaS in 2026](https://www.agentickit.co/blog/best-cursor-skills-stripe-billing-saas) - Signature verification, idempotent webhooks, subscription state sync, dunning, and billing portal UI — 8 Cursor skill/rule combos for the Stripe billing failures that actually ship, with a copy-paste rule snippet.
- [Your Vibe-Coded App Is Leaking API Keys: The 60-Second Check](https://www.agentickit.co/blog/vibe-coding-exposed-api-keys) - Hacking a vibe-coded app is a Network tab and one search for "sk-". Run the check on your own site, see what an exposed key really costs, and fix the actual gap: nothing is reviewing what the AI ships.
- [NVIDIA Nemotron 3 Ultra in OpenCode: Benchmarks and a Free Setup Guide](https://www.agentickit.co/blog/nemotron-3-ultra-opencode-guide) - No, it did not kill Claude Code. But a 550B open model with 1M context, free via NVIDIA NIM, running in OpenCode is a real stack. Benchmarks, config, and troubleshooting.
- [How Solo Founders Can Use LinkedIn Outreach to Get Their First 100 Customers](https://www.agentickit.co/blog/linkedin-outreach-first-100-customers) - Distribution beats another build week. A practical LinkedIn outreach system for solo founders: ICP, prospecting, personalization, automation, and follow-up.
- [Cursor Rules for SaaS Founders: A Production Setup That Agents Actually Follow](https://www.agentickit.co/blog/cursor-rules-saas-founders) - Four focused .mdc files beat one manifesto. A production Cursor rules stack for multi-tenant SaaS, with copy-paste API, database, and tenant examples.
- [My Complete Cursor Workflow for Shipping a SaaS Feature](https://www.agentickit.co/blog/cursor-workflow-shipping-saas) - Spec, contract, schema, API, UI, tests, review, security, then a ship gate. A complete Cursor workflow walked through a teammate-invite feature.
- [Cursor vs Claude Code for Solo SaaS Founders: 2026 Head-to-Head (With Numbers)](https://www.agentickit.co/blog/claude-code-vs-cursor-2026-saas-founders) - Not another benchmark table. A cost-per-feature-shipped breakdown of Claude Code's terminal loop vs Cursor's inline diff, which jobs go to which tool, and what rules/skills actually port between them.
- [Cursor vs Claude Code: Context Windows, Hooks, and Subagent Enforcement](https://www.agentickit.co/blog/cursor-vs-claude-code-context-hooks-enforcement) - Not the cost table or the job list — this one is about enforcement. Why Claude Code's hooks can block a bad commit and Cursor's rules can't, and what subagents do to context-window pressure on long tasks.
- [Cursor vs Claude Code: What I Actually Use to Ship SaaS](https://www.agentickit.co/blog/cursor-vs-claude-code) - Not a feature matrix. A job-based comparison of two harnesses, with an honest Cursor-native bias and a week of SaaS work mapped to each.
- [How I Structured 34 AI Agents for AgenticKit](https://www.agentickit.co/blog/how-i-structured-34-ai-agents) - A pile of agent files is not a team. The org chart, routing rules, and why /ship adopts roles in one session instead of forking 34 processes.
- [Cursor Memory Bank: A Practical Setup That Stays Current](https://www.agentickit.co/blog/cursor-memory-bank-setup) - Community memory banks go stale. One living CURSOR.md, glob rules, and a sessionStart hook keep the stack honest without a six-file religion.
- [I Rebuilt My SaaS Workflow Around AI Agents](https://www.agentickit.co/blog/rebuilt-saas-workflow-ai-agents) - Before: one chat for billing, headlines, and a launch tweet. After: kits on disk, two loops, and a human in the review seat. What got slower included.
- [35 Cursor Slash Commands I Actually Use to Ship SaaS](https://www.agentickit.co/blog/cursor-slash-commands-i-use) - All 35 AgenticKit commands grouped by job: build, verify, operate, grow. When to use each, when not to, and a week mapped to slashes.
- [What Went Wrong With My First AI Coding Workflow](https://www.agentickit.co/blog/first-ai-coding-workflow-mistakes) - Unbounded chats, a novel instead of rules, self-review, tenant leaks, and marketing in the SQL thread. Each failure mapped to a guardrail.
- [Cursor for Solo Founders: From Idea to Production](https://www.agentickit.co/blog/cursor-solo-founders-idea-to-production) - Positioning, one paid vertical slice, a weekly ship and launch loop, then production checks. Cursor is the workbench, not the company.
- [Building AgenticKit: Architecture, Workflows, and Lessons](https://www.agentickit.co/blog/building-agentickit-architecture-lessons) - A Cursor kit is a context graph: agents, skills, commands, rules, and a living CURSOR.md. How it is packaged, and what I would change.
- [5 Cursor Features That Will Save Your Time Every Day](https://www.agentickit.co/blog/5-cursor-features) - Boost your daily coding speed with inline prompts, Agent mode, codebase indexing, automated PR reviews, and slash workflows in Cursor.
- [We Audited Our Own Search Console Data: 4 Page Fixes That Actually Moved Position](https://www.agentickit.co/blog/search-console-page-audit-fixes) - Impressions with 0% CTR and a position-29 near-miss are the highest-leverage rows in Search Console. The 4 fixes we shipped, with the exact GSC numbers behind each.
- [Claude Code Hooks for Founders: 7 Ready-to-Paste JSON Snippets to Stop Dumb Mistakes](https://www.agentickit.co/blog/claude-code-hooks-examples) - A hook can block an action; a rule is just advice the model reads. 7 copy-paste .claude/settings.json snippets for the mistakes that actually cost founders time.
- [15 CLAUDE.md Rules Solo SaaS Founders Actually Need in 2026](https://www.agentickit.co/blog/claude-md-rules-solo-saas-founders-2026) - Generic CLAUDE.md advice ('write clean code') changes nothing. The rules that work are pulled from an actual correction Claude needed — 15 of them, grouped by scope, how to write one from a mistake, and solo-founder framing, plus a template.
- [15 Real CLAUDE.md Files From Production SaaS Repos (What Actually Works)](https://www.agentickit.co/blog/claude-md-examples-production-saas) - Every CLAUDE.md post right now is generic advice with zero real files to copy. Here are three annotated CLAUDE.md files — Next.js + Stripe, Postgres/Drizzle, and a monorepo — with a line-by-line note on why each line earns its place.
- [Claude Code CLAUDE.md Length Audit: The 200-Line Rule Explained](https://www.agentickit.co/blog/claude-md-200-line-rule) - 'Keep CLAUDE.md under 200 lines' shows up in guide after guide with no source. It isn't in Anthropic's docs — here's what is, the deletion test that actually predicts bloat, and a before/after trim using the CLAUDE.md Linter.
- [20 Best Claude Code Skills for Solo SaaS Founders in 2026 (Ranked & Tested)](https://www.agentickit.co/blog/best-claude-code-skills-2026) - Most 'best Claude Code skills' posts list names from a directory. We installed AgenticKit's catalog into .claude/skills/ and ran one real task per skill — here's the ranked 20, with token cost and the primitives primer most roundups skip.
- [I Audited My Own Claude Code Subagent Setup With Search Console Data — Here's What Was Broken](https://www.agentickit.co/blog/subagent-setup-audit-fixes) - Two agents claiming the same task, a reviewer with edit access it shouldn't have, and a subagent that never needed its own context window. The actual audit.
- [20 Best Claude Code Subagents for SaaS Founders in 2026 (Ranked)](https://www.agentickit.co/blog/best-claude-code-subagents-2026) - code-reviewer without edit access, security-auditor that can't patch its own findings, billing-engineer scoped to Stripe only — 20 real subagent roles ranked, with the tool grant each one actually needs.
- [Claude Code Skills vs. Subagents vs. Commands: A Decision Guide for SaaS Founders](https://www.agentickit.co/blog/claude-code-skills-vs-subagents-vs-commands) - Commands, subagents, and skills solve different problems, but roundups rarely explain which one to write for a given task. Here's a 3-question decision tree with real SaaS-founder examples and the mistakes that tank response quality.
- [Claude Code Agent Teams vs. Subagents: What Actually Changed in 2026](https://www.agentickit.co/blog/claude-code-agent-teams-vs-subagents) - Blogs keep using 'subagents' and 'Agent Teams' as synonyms. They aren't — one is a delegation unit, the other is an arrangement of units. Here's the actual distinction, with a decision table.
- [Claude Code Rules for SaaS Billing & Stripe Webhooks: Setup Guide 2026](https://www.agentickit.co/blog/claude-code-stripe-webhooks-setup) - A .claude/skills/stripe-billing-webhooks/SKILL.md that forces raw-body signature verification, idempotent event handling, and subscription state sourced only from Stripe's own events — annotated, with the Cursor rules equivalent.
- [The Title-Tag-First Rule: A Second Search Console Audit Pass (2026)](https://www.agentickit.co/blog/snippet-rewrite-playbook-second-audit) - Running the same Search Console audit a second time turned one-off fixes into a repeatable rule: fix the title tag and meta first on a near-miss, the content only on a deep-position page.
- [Search Console Audit, Pass Five: Why Page-1 CTR and Page-3 Relevance Need Different Fixes](https://www.agentickit.co/blog/search-console-audit-pass-five) - Position 9 with 0% CTR is not the same problem as position 67 with 0% CTR. The fifth pass in this series adds the rule for fixing each, plus a one-page fix for a query split across two rankings.
- [Search Console Audit, Pass Six: The Fixes From Pass Five Hadn't Registered Yet](https://www.agentickit.co/blog/search-console-audit-pass-six) - Pulling the same rows again two days after pass five shipped is a lesson in patience: nothing had moved yet. The real work this pass was a cannibalization fix between /commands and its own blog post, and a 2026 near-miss on the AGENTS.md vs CLAUDE.md comparison.
- [12 Best Claude Code Plugins in 2026 (Marketplaces, Install Commands, What to Skip)](https://www.agentickit.co/blog/best-claude-code-plugins-2026) - Plugins bundle commands, skills, subagents, and hooks into one /plugin install — ranked across 12 picks, with the audit steps to run before trusting a community marketplace with your repo.
- [Claude Code Plugin Marketplace Governance: What Solo Founders Need to Know in 2026](https://www.agentickit.co/blog/claude-code-plugin-marketplace-governance-2026) - No binary signing for Claude plugins yet means trust comes down to which marketplace you install from and how carefully you read the source — here's the 5-minute checklist and when a managed-settings allowlist is actually overkill.
- [We Audited 4 Pages for Ecosystem Framing: The Cursor-vs-Claude-Code Fix That Mattered](https://www.agentickit.co/blog/cursor-claude-code-framing-audit-fixes) - Position 66.7 with 0% CTR on 'cursor skills' and a 3.8-position branded query still getting 0% CTR: the same root cause, two symptoms. The 4 fixes we shipped.
- [15 Claude Code Hooks That Actually Prevent Mistakes (With Copy-Paste Configs)](https://www.agentickit.co/blog/15-claude-code-hooks-examples) - Most guides to this topic explain the 12 lifecycle events like a reference manual and skip the part that matters: what exit code 2 actually does. 15 copy-paste hooks across six categories, led with the exit-code mechanics.
- [Claude Code Hooks Cookbook: 12 Automations Solo Founders Ship](https://www.agentickit.co/blog/claude-code-hooks-cookbook-12-automations) - Organized by job, not by event: 3 formatting hooks, 3 safety guardrails, 3 notification automations, 3 testing/CI gates, and one settings.json template that merges a sane default from each category.
- [Claude Code Hooks in Production: The Advisory vs Deterministic Rule Nobody Explains](https://www.agentickit.co/blog/claude-code-hooks-advisory-vs-deterministic) - 2026 hooks guides explain PreToolUse and PostToolUse but skip the concept that actually decides whether a rule holds: CLAUDE.md is advisory text the model can deprioritize, a hook is a command the harness runs every time, no exceptions.
- [20 Claude Code Hooks Examples That Actually Ship in Production (2026)](https://www.agentickit.co/blog/20-claude-code-hooks-examples-production) - Most hooks lists repeat the same 5 configs in a different order. This one ranks all 20 by ROI, covers async:true hooks for logging without adding latency, and gives the tool_input schema table so you can write your own.
- [Search Console Audit, Pass Four: A New Row and Three We're Watching](https://www.agentickit.co/blog/search-console-audit-pass-four) - Position 29.1 on "cursor rules skills docs" is a framing problem, not a ranking one. The one new fix we shipped, and why we didn't touch three repeat rows.
- [20 Claude Code Agent Teams for Marketing, SEO, and Growth (2026 Ranked)](https://www.agentickit.co/blog/claude-code-marketing-agent-team-2026) - seo-specialist that never writes the article, growth-strategist that never touches code, customer-marketer that can't invent a quote — 20 real marketing agent roles ranked by how a solo founder actually uses them.
- [Claude Code Subagents Best Practices: Choosing Your Delegation Model](https://www.agentickit.co/blog/claude-code-subagents-best-practices) - "Use proactively" in a description field does more for auto-delegation than a longer system prompt ever will. Five practices for subagent design, what a subagent actually inherits from CLAUDE.md, and the one mistake — doing too much — that breaks routing for every agent after it.
