Inside davila7's claude-code-templates Plugin

What is claude-code-templates by davila7?

davila7/claude-code-templates is one of the largest single plugins in Flowy's index. Flowy counts 500 components for it: 151 commands, 165 agents, 29 MCP servers, and 155 skills, indexed from the public GitHub repository, which carries 30,290 stars. Flowy indexes that repository; it does not own, bundle, or host any part of it, so use the listing page linked at the top of this guide for the current install path rather than anything printed here.

The repository reads less like one focused tool and more like a library built for many separate jobs: running a Google Workspace account through slash commands, scaffolding a Next.js app, reviewing Azure infrastructure, fine-tuning a language model, or writing a resume. The sections below group the roster by what each family is actually for, drawn from the components' own descriptions rather than guessed from a name.

Why do so many commands automate Google Workspace?

More than half of the 151 commands exist to drive a Google Workspace account. gws-shared describes itself as shared authentication, flag, and output-formatting patterns for a companion command-line tool the rest of the family calls into, and dozens of gws-prefixed commands each wrap one service: gws-gmail for mail, gws-calendar for events, gws-drive for files, gws-docs and gws-slides for documents and decks, gws-vault for eDiscovery holds, gws-admin for users and devices, and gws-modelarmor for filtering content through Google's Model Armor safety service, among many more covering Chat, Sheets, Keep, Groups, and Workspace Events.

A second layer of recipe-prefixed commands chains those primitives into one finished task. recipe-generate-report-from-sheet reads a Sheet and writes a formatted Docs report; recipe-share-event-materials shares Drive files with everyone on a Calendar invite; dozens more cover email, scheduling, and file organization the same way. A third, small layer of eight persona-prefixed commands bundles recipes into a role instead of a single action: persona-hr-coordinator, persona-it-admin, persona-project-manager, persona-customer-support, persona-event-coordinator, persona-researcher, persona-team-lead, and persona-content-creator.

What does the rest of the command list cover?

Outside Google Workspace, the commands read like a working developer's toolbox. Six nextjs-prefixed commands, including nextjs-scaffold and nextjs-migration-helper, cover scaffolding, Pages Router migration, performance and bundle audits, component generation, and middleware. Eight supabase-prefixed commands, including supabase-schema-sync and supabase-security-audit, cover schema sync, migrations, security audits, type generation, performance tuning, realtime monitoring, backups, and data exploration. Git and release flow gets five worktree commands (worktree-init, worktree-check, worktree-deliver, worktree-cleanup, and create-worktrees) plus release, hotfix, branch-cleanup, commit, and create-pull-request. Smaller sets cover Vercel deployment (vercel-deploy-optimize, vercel-env-sync, vercel-edge-function), cross-posting a blog post to four platforms (publisher-x, publisher-devto, publisher-medium, publisher-all), and, unexpectedly for a coding plugin, group food orders (doordash-budget, doordash-playbook, doordash-lunch).

What is in the 165-agent roster, and who is it really for?

The general engineering set includes debugger, refactoring-specialist, code-architect, a paired tdd-red and tdd-green, and test-runner. An infrastructure set centers on ten Azure agents (azure-infra-engineer, azure-iac-generator, bicep-plan, and more) plus kubernetes-specialist and sre-engineer. A database set runs five Neon specialists (neon-database-architect, neon-auth-specialist, neon-migration-specialist, and two more) alongside postgres-pro and nosql-specialist. A research set, led by research-orchestrator, moves a query through query-clarifier, specialist researchers, fact-checker, and research-synthesizer to a final report-generator.

Not every agent is aimed at your codebase, though. About ten, by their own descriptions, exist to run davila7's own contribution pipeline: catalog-generator regenerates the project's component catalog, component-reviewer checks new components before they merge, component-migrator and component-improver pull in and patch components from other repositories, command-expert, agent-expert, and mcp-expert draft new commands, agents, and MCP servers for this same project, and linear-tracker and blog-writer manage its issue tracker and blog.

What do the skills and MCP servers add?

The 155 skills lean hardest into machine learning engineering, close to a third of the total: fine-tuning (fine-tuning-peft, fine-tuning-axolotl, fine-tuning-unsloth), distributed training (distributed-training-deepspeed, distributed-training-pytorch-fsdp), inference serving (inference-serving-vllm, inference-serving-sglang), and quantization (optimization-awq, optimization-gptq, optimization-gguf). Two smaller clusters sit next to it: five conversion-rate skills (page-cro, form-cro, popup-cro, paywall-upgrade-cro, onboarding-cro) and thirteen career-document skills (resume-bullet-writer, cover-letter-generator, linkedin-profile-optimizer, executive-resume-writer, and more). The 29 MCP servers skew toward developer infrastructure: github, docker, database servers for postgres, postgresql, mysql, and neon, language documentation servers such as python-sdk, typescript-sdk, and rust-docs, plus slack and puppeteer.

Who should install this?

Given the range, from Google Workspace automation to LLM quantization to resume writing, this plugin works better as a searchable library than a single workflow. Install it when one specific, well-described command, agent, MCP server, or skill matches a job you already have, and check the listing page linked at the top of this guide for the current install path before adding any of it to a project.