Food_For_

Pulls version-correct docs for whatever library you're using straight into the prompt, so Claude stops inventing API methods that were removed two majors ago.

ForAnyone whose stack moves faster than the model's training cutoff.

Drives a real browser through accessibility snapshots instead of screenshots — clicks, forms, navigation, assertions, all deterministic and no vision model needed.

ForFrontend devs who want Claude to check its own UI work before claiming it's done.

Hands Claude the actual DevTools: performance traces, network waterfalls, and console errors with source-mapped stack traces from a live Chrome.

ForAnyone debugging a slow page or a bug that only reproduces in the browser.

Repos, issues, pull requests, code search, and Actions runs — read and write, across every repo you can already see.

ForTeams who want triage, review, and CI babysitting in the same session as the code.

Scoped read and write access to directories you name explicitly — the safe way to give a desktop client hands on local files.

ForClaude Desktop users; anyone who wants file access limited to exactly one folder.

Gives the model a scratchpad for multi-step problems it can revise mid-flight — branch a thought, walk it back, try the other path.

ForArchitecture calls and gnarly debugging where the first answer is usually wrong.

A local knowledge graph of entities and relations that survives between sessions, so context you established last week is still there today.

ForLong-running projects where re-explaining the codebase every session is the tax you're tired of paying.

Read and file issues, move them through cycles, and pull project state — so the ticket gets updated by whoever actually did the work.

ForProduct and eng teams who live in Linear and want the loop closed automatically.

Search, read, and write pages and databases — turn a messy thread into a spec, or pull the spec you already wrote into the build.

ForTeams whose real documentation lives in Notion, not the repo.

Pulls production errors with full stack traces and breadcrumbs, so you go from alert to root cause without opening another tab.

ForOn-call engineers debugging something that already broke for real users.

Query the database, run migrations, deploy edge functions, generate types, and read logs and security advisories against your live project.

ForAnyone shipping on Supabase who wants schema changes and app code to move together.

Reads real design context out of a file — variables, components, layout, Code Connect mappings — instead of guessing from a screenshot.

ForFrontend teams implementing designs that must match the design system exactly.

Searches Vercel docs, inspects projects and deployments, reads build and runtime logs, and queries Web Analytics.

ForAnyone whose deploy just failed and doesn't want to go read the dashboard.

A family of OAuth-secured servers covering Workers, DNS, observability, and docs search across your Cloudflare account.

ForTeams running edge infra who want config and debugging in one place.

Inspect customers, subscriptions, and payment intents, and search Stripe's docs while wiring up a checkout that has to work the first time.

ForDevelopers building billing who need to see what the API actually returned.

Neural web search built for agents — semantic queries, full page contents, plus company and people research in a single call.

ForResearch-heavy work where keyword search returns SEO sludge.

Turns any site — including JS-rendered ones — into clean markdown, and crawls whole doc sets. The v2 endpoint works with no key at all.

ForAnyone ingesting documentation, competitors, or a site with no usable API.

281k+ real screens from shipped products, searchable — so design decisions come from what actually works rather than the model's median taste.

ForFounders and teams building onboarding, pricing, or dashboards without a designer.

Search models, datasets, Spaces, and papers, and call inference endpoints — the whole Hub as tools.

ForML engineers picking a model or hunting for a dataset that already exists.

Semantic code search over millions of lines using a vector index, so Claude finds the right file without grepping through your monorepo.

ForTeams whose codebase is far too large to fit in any context window.

Headless browsers in the cloud with Stagehand's natural-language actions, session recording, and stealth mode — no local Chrome required.

ForAgents that need to browse from CI or a server, not your laptop.

Jira and Confluence together — search issues, file and transition tickets, and read or write Confluence pages from one endpoint.

ForEnterprise teams whose planning lives in Jira and whose docs live in Confluence.

Serverless Postgres with real branching — spin up a database branch, run the migration against it, and throw it away if it was wrong.

ForDevelopers who want to test destructive migrations without holding their breath.

Query product analytics, build insights, read session recordings, and manage feature flags — ask what users actually did, in English.

ForProduct teams who want the funnel answer without writing the query.

Search dashboards, run PromQL and Loki queries, and pull incidents — so debugging starts from the metrics rather than a screenshot of them.

ForSREs and on-call engineers working from Grafana dashboards.

Pull metrics, logs, traces, monitors, and incidents into the session so you can correlate a spike with the deploy that caused it.

ForTeams running Datadog as their primary observability stack.

Create and configure sites, manage env vars and deploys, and read build logs when the build breaks for reasons the log doesn't explain well.

ForTeams shipping frontends and functions on Netlify.

Manage Postgres databases and run schema migrations through Prisma's own tooling, with the schema as the source of truth.

ForTypeScript teams using Prisma as their ORM.

Explore collections, run aggregation pipelines, manage indexes, and inspect Atlas clusters without leaving the session.

ForTeams on MongoDB or Atlas doing schema work and query tuning.

More than a query runner — it does index tuning, EXPLAIN plan analysis, and health checks, with a read-only mode you should probably use.

ForAnyone whose Postgres query got slow and who doesn't want to guess at indexes.

Inspect and manipulate keys, streams, and vector indexes — useful for debugging a cache that is confidently serving the wrong thing.

ForBackend developers debugging cache and queue state.

Query your indices in natural language, inspect mappings, and iterate on search relevance against real data.

ForTeams running Elasticsearch who tune relevance by trial and error.

A vector database as a memory layer — store and semantically retrieve snippets, so the agent recalls by meaning rather than exact match.

ForAnyone building RAG or giving an agent long-term recall.

Read, create, and update records and schema across your bases — the practical way to let Claude drive a spreadsheet-shaped database.

ForOps and non-engineering teams whose system of record is Airtable.

Searches the provider registry and module docs so generated HCL uses arguments that actually exist in the provider version you pinned.

ForPlatform engineers writing Terraform against unfamiliar providers.

Runs MCP servers in containers behind one gateway, with secrets held outside the model's reach — the sane way to run a dozen servers at once.

ForTeams who want MCP sandboxed rather than running npx straight on the host.

A suite covering documentation, CDK, cost analysis, Lambda, and more, each scoped to one job rather than one giant AWS tool.

ForCloud engineers building on AWS who want grounded, current service docs.

Independent web index with news, image, and local search — a search source that isn't reselling someone else's results.

ForAnyone who wants web search with a privacy stance and a free API tier.

Opens up 6,000+ prebuilt scrapers — social, maps, ecommerce, search — so you skip writing the scraper someone already wrote.

ForAnyone extracting structured data from sites that fight scrapers.

Ask questions about any public GitHub repo and get answers grounded in an indexed wiki of that codebase. No key, no signup.

ForAnyone adopting an unfamiliar open-source dependency.

Static analysis that finds real vulnerability patterns in the diff — injection, authz gaps, leaked secrets — and explains why each one matters.

ForAnyone who wants a security pass before the PR, not after the incident.

Reaches roughly 8,000 apps through actions you pre-authorize, which covers the long tail no dedicated MCP server exists for.

ForOps teams automating across SaaS tools that will never ship their own server.

Read and update contacts, companies, and deals, and query the CRM in plain language instead of building a report view for it.

ForSales and marketing teams running pipeline out of HubSpot.

Search conversations and tickets, so you can trace a bug report back through what the customer actually said.

ForSupport and product teams mining conversations for patterns.

Create and update tasks, projects, and portfolios, and pull status without opening the board and scrolling.

ForTeams coordinating cross-functional work in Asana.

Read and create boards, sticky notes, and diagrams — turn a messy workshop board into a structured doc, or the reverse.

ForTeams who run planning on a whiteboard and need it written up afterward.

Generate and edit designs, resize for each channel, and pull assets out of your brand kit.

ForMarketers producing volume creative without a designer in the loop.

Manage CMS collections, items, and site publishing — bulk content edits without clicking through the designer.

ForMarketing sites running on Webflow CMS.

Create invoices, manage orders and subscriptions, and pull transaction reports through PayPal's official server.

ForMerchants and developers building on PayPal.

Payments, catalog, inventory, orders, and customers across the Square API surface.

ForRetail and restaurant businesses running on Square.

Query and mutate structured content with GROQ, and manage schemas and releases in a content lake built for editing by API.

ForTeams running headless content who want bulk edits done properly.

Manage site content, products, bookings, and business data across Wix sites you own.

ForSmall businesses and agencies running client sites on Wix.

Pulls feature requests, insights, and roadmap state so you can ask what customers keep asking for instead of re-reading the board.

ForPMs running discovery and prioritisation on real customer input.

Ask behavioural questions in plain language — funnels, retention, cohorts — and get the chart without building it first.

ForPMs who need the number now and not after a data request.

Event analytics on demand: query what happened, segment it, and check whether the thing you shipped moved anything.

ForProduct teams validating a launch in the first 48 hours.

Generates and iterates real UI components you can look at and edit — prototyping that produces something clickable rather than a description.

ForPMs and founders who need a mock to react to before involving design.

Sketches architecture diagrams, flows, and wireframes in that hand-drawn style that reads as a draft, so people critique the idea and not the pixels.

ForAnyone whiteboarding a system or a user flow. No key required.

Validates Mermaid syntax and renders real SVG, which catches the broken diagram before it lands in your doc. Works with no key.

ForAnyone putting diagrams in specs, READMEs, or PRs.

A CRM whose data model you actually define — query and update records, lists, notes, and calls without a rigid schema fighting you.

ForFounders and operators running a modern, custom-shaped pipeline.

Stories, epics, and iterations for teams who found Jira heavy — file and move work without the ceremony.

ForSmall product teams who want tracking, not process.

Search and edit your developer docs in place, so the doc gets updated in the same session as the change it describes.

ForTeams whose docs drift out of date within a week of every release.

Reaches your collections and environments to run requests and inspect responses — checking an API against reality, not its docs.

ForDevelopers integrating an API that behaves differently than documented.

Scores npm and PyPI packages for supply-chain risk — install scripts, obfuscated code, network calls — before you add the dependency.

ForAnyone letting an agent add packages to their project. No key required.

Manage users, sessions, and organisations, and get auth wired correctly the first time instead of after a security review.

ForBuilders adding sign-in to a product without rolling their own.

SSO, SCIM directory sync, and audit logs — the enterprise checklist that blocks your first big deal.

ForStartups moving upmarket and hitting procurement for the first time.

Send and inspect transactional email, manage domains, and check deliverability from the session you're building in.

ForDevelopers wiring up email that has to land in the inbox.

Product email built for SaaS — contacts, lists, and lifecycle campaigns without a marketing automation suite bolted on.

ForSmall teams running onboarding and lifecycle email themselves.

Notification infrastructure — in-app, email, push, and Slack from one workflow, with preferences and batching handled for you.

ForBuilders who realised notifications are a whole product surface.

Serverless Redis, Vector, and QStash priced per request — inspect keys, manage queues, and query vector indexes.

ForBuilders who need a cache or queue without running a server for it.

Create and manage services, databases, and env groups, and read deploy logs when a build fails for an unhelpful reason.

ForSmall teams who want a Heroku-shaped deploy that still scales.

Query logs and events at scale in APL without the per-gigabyte anxiety of the incumbents.

ForTeams who want to keep all their logs and still afford to search them.

Build and inspect internal tools and workflows — the admin panel nobody wants to hand-write.

ForOps and eng teams shipping internal CRUD without a sprint for it.

Visual-editor headless CMS — manage stories, components, and releases, so editors keep their preview and you keep your API.

ForTeams whose marketers need to edit without filing a ticket.

Tracks SOC 2 and ISO controls, evidence, and failing checks — compliance state as data instead of a spreadsheet someone owns.

ForStartups going through their first audit.

Newsletter platform with real analytics — drafts and publishes posts, segments subscribers, and reads what each send actually did.

ForFounders and creators running a newsletter as a growth channel.

The one everybody recommends. A full methodology — brainstorm, write a spec, branch into git worktrees, then drive it with TDD — as composable skills that fire on their own.

ForTeams who want Claude to follow a process instead of sprinting straight at the code.

docx, pptx, xlsx, and pdf — creates and edits real Office and PDF files with working formulas, tracked changes, layouts and page numbers, not markdown pretending to be a report.

ForAnyone whose deliverable has to open in Word, Excel, PowerPoint, or Acrobat.

Scaffolds a new skill, then evaluates and tunes its description so it actually triggers when it should — the hardest part of writing one.

ForAnyone who has written a skill that never fires.

Walks you from an external API to a well-shaped MCP server — tool design, schemas, error handling, and the evals to prove it works.

ForDevelopers wrapping an internal API so their agent can use it.

Pushes generated UI away from the default purple-gradient-on-white look toward production-grade interfaces with real typography and a considered palette.

ForAnyone tired of every AI-built page looking like the same AI built it.

Drives a real browser against your local app to verify the feature works — clicks the flow, reads the console, catches what unit tests never do.

ForDevelopers who want proof, not a confident summary.

Builds richer interactive artifacts — multi-component React and Tailwind pages that hold state — rather than one static block of HTML.

ForAnyone shipping interactive tools and dashboards as artifacts.

Multi-artboard visual design on one pan-and-zoom canvas — mockups, screen flows, posters, one-pagers you can then nudge by hand.

ForProduct people who need to show a layout, not describe one.

Generates and applies coherent color and type themes across artifacts, so a set of documents reads as one system instead of five different days.

ForAnyone producing a run of decks, docs, or pages that need to match.

Teaches Claude your brand's voice, color, and type rules once, then holds every output to them.

ForMarketing and design teams shipping anything customer-facing.

Drafts announcements, project updates, and incident notes in the register internal writing actually uses — short, specific, no throat-clearing.

ForManagers and leads who write the same five formats every week.

Structures a genuine back-and-forth on a long document — outline, draft, critique, revise — instead of dumping 4,000 words and calling it done.

ForPeople writing specs, strategy docs, or anything that needs more than one pass.

Current model IDs, pricing, parameters, tool-use and caching patterns — so Claude stops writing code against an API shape from last year.

ForDevelopers building on the Claude API or the Agent SDK.

Generative visual work done properly — flow fields, particle systems, and tilings rendered on Canvas or WebGL rather than hand-written SVG paths.

ForDesigners and creative coders making something to look at.

Builds animated GIFs sized and compressed to actually upload to Slack, which is the part that normally goes wrong.

ForAnyone whose team communicates primarily in reaction GIFs.

Persistent memory that carries decisions and context across sessions, so /clear stops costing you an hour of re-explaining.

ForPower users on projects that span weeks.

Roughly 70 React and Next.js rules on performance, hooks, accessibility, and TypeScript, applied as a review pass after you edit components.

ForReact teams who want the code review to happen before the pull request.

Point it at a documentation site and it crawls, structures, and packages the whole thing into a reusable skill.

ForAnyone working against a library the model doesn't know well.

Keeps the plan in markdown on disk rather than in context, so a crash, a compaction, or a /clear doesn't lose the thread.

ForAnyone running long multi-step builds that outlive one context window.

Interrogates your plan before a line gets written — asks the questions that surface the requirement you forgot to mention.

ForAnyone who has built the wrong thing very efficiently.

Boots and drives the iOS Simulator — launch the build, tap through it, screenshot the result, read the logs.

ForiOS developers who want Claude to verify the screen, not just compile it.

SwiftUI layout and state guidance with a review pass that catches the patterns that compile fine but feel wrong on device.

ForiOS developers building UI in SwiftUI.

Builds real interactive D3 charts — proper scales, axes, transitions, and tooltips — instead of a static SVG that only looks like a chart.

ForAnalysts and data teams who need the visualization to be explorable.

Runs ffuf properly for content discovery and parameter fuzzing, with wordlist and filtering choices that cut the false positives.

ForSecurity engineers and bug bounty hunters doing authorized testing.

One skill, 1000+ services — send the email, file the issue, post the message, update the row, across Gmail, Slack, GitHub, Notion and the rest.

ForAnyone who wants Claude to finish the task, not just draft it.

Four phases before any fix: reproduce it, isolate it, find the root cause, then change one thing. Stops the guess-and-check spiral that breaks two more things.

ForAnyone whose last debugging session ended in five speculative edits.

Enforces a real RED-GREEN-REFACTOR loop — the test must fail first, which is the step agents skip when left alone.

ForTeams who want tests that actually prove something.

Gives each task its own worktree so parallel agents stop fighting over one working directory and your main branch stays clean.

ForAnyone running more than one agent at a time.

Pulls the actual requirements out of you through questions before any code exists, then writes them down as a spec you can argue with.

ForAnyone who starts with a vague idea and wants it sharp before building.

Blocks the victory lap. Forces a check that the thing actually runs and the tests actually pass before anything gets called done.

ForAnyone who has been told a build was finished when it did not compile.

Splits a large build across subagents with clean handoffs, so each one works in a fresh context instead of one session degrading over hours.

ForTeams tackling migrations and refactors too big for a single context window.

Packages a change into a review request a human can actually read — the diff, the reasoning, and what to look at hardest.

ForAnyone whose PR descriptions say "various fixes".

Supabase's own guidance on RLS policies, migrations, and auth patterns — the security rules that are easy to get quietly wrong.

ForAnyone building on Supabase who would rather not ship an open bucket.

119+ slash commands covering code review, security audits, architecture analysis, and release chores, organized so you can find them.

ForDevelopers who keep retyping the same long prompts.

A large catalog of ready-made agents, commands, hooks, and settings you can drop into a project instead of configuring from scratch.

ForAnyone setting up a new repo who wants sensible defaults immediately.

68 product-management skills — discovery, prioritization, PRDs, roadmaps, and metrics — with output samples and enforced contracts.

ForPMs who want a rigorous process rather than a blank doc.

Eleven SEO and content skills covering keyword research, on-page structure, technical audits, and briefs that writers can follow.

ForContent and growth teams doing SEO without an agency.

Teaches the Playwright patterns worth using — resilient selectors, proper waits, and fixtures — so end-to-end tests stop being flaky.

ForTeams whose E2E suite fails randomly and gets ignored.

Captures a whole scrolling page through Chrome DevTools with no dependencies to install.

ForAnyone documenting UI or filing visual bug reports.

Uses combinatorial test design to cover every pair of parameters in a fraction of the cases exhaustive testing would need.

ForQA engineers testing features with many configuration permutations.

AWS architecture and CDK patterns with the service-specific details that generic cloud advice always gets wrong.

ForCloud engineers writing infrastructure as code on AWS.

Turns markdown or a long chat into a properly structured EPUB with chapters and metadata that e-readers respect.

ForWriters and researchers who read long-form on a Kindle or Kobo.

Strips replies down to the essentials. Fewer output tokens, less preamble, lower bill.

ForHeavy users watching their token spend.

The long-running community index of commands, workflows, hooks, and configuration — where to look when you suspect someone already solved it.

ForAnyone assembling their own Claude Code setup.

50+ skills that were actually checked before being listed, with categories and install notes — the useful kind of awesome-list.

ForAnyone browsing for skills without wanting to test every one.

A very large bundle — 135 agents, 35 skills, 42 commands, hooks and templates — for raiding rather than installing whole.

ForPower users assembling a custom setup from parts.