11ai

Reusable AI-agent skills for building chat products, editorial blogs, and repo-driven project sites; automating agent work; benchmarking coding models; reverse engineering and maintaining codebases; integrating application services; running repository tasks through disciplined Git workflows; cleaning up local development environments; and running deep iterative project audits.

Project site: https://ai.rj11.io/

Each skill is a self-contained directory led by a SKILL.md file. Keep the whole directory together so its references, scripts, assets, and agent metadata remain available.

Install

Install the collection from GitHub with the skills CLI:

npx skills add rj11io/11ai --full-depth

The CLI runs through npx, discovers the repository's skills, and configures the skills you select for your agent. --full-depth is required because this collection organizes skills into twenty workflow plugins instead of one top-level skills/ directory.

The repository follows the open Agent Skills format. Every skill also includes Codex UI metadata in agents/openai.yaml, and every plugin is packaged for Claude Code. To add the Claude marketplace, run:

/plugin marketplace add rj11io/11ai

Then install the plugin you need from the marketplace UI or with /plugin install <plugin>@11ai.

After installation, ask your agent to use a skill by name. For example:

Use 11ai-ai-chat-stack to add a complete AI chat surface to this application.

Package-only installation

Install the npm package directly when you need a versioned copy of the raw files rather than agent configuration:

npm install --save-dev @rj11io/11ai

The files are installed at node_modules/@rj11io/11ai/v0/plugins/. You can also clone this repository and use v0/plugins directly.

Skill catalog

The repository currently contains 155 skills in 20 plugins.

| Plugin | Skills | Use them for | | --- | ---: | --- | | Agent automation | 8 | Scheduled autonomous work using the Ledger + Conductor + Routine pattern | | AI chat | 9 | End-to-end chat, sessions, tools, models, providers, UI, and messaging-platform extensions | | Ant Design operations | 10 | Common Ant Design React setup, layouts, forms, tables, navigation, overlays, display, theming, reference, and troubleshooting | | AWS operations | 11 | Common AWS CLI operations for account context, S3, EC2, Lambda, ECS, ECR, CloudWatch, IAM, CloudFormation, reference, and troubleshooting | | Benchmarks | 19 | Running resumable soft/final lifecycles; creating, auditing, AI/human judging, accounting, reviewing, synchronizing, and publishing benchmarks | | Blog builder | 8 | Building file-backed editorial blogs with composable CMS, author, Markdown, content, navigation, and UI skills | | Cleanup | 5 | Safely finding and removing abandoned local resources | | Codebase | 3 | Playwright setup, npm publishing, and automated releases | | Docker operations | 10 | Common Docker CLI and Compose operations, safe cleanup, and troubleshooting | | Git operations | 12 | Modular, safety-first skills for common Git operations, repository workflows, learning, and troubleshooting | | Integrations | 3 | Adding Convex, WorkOS AuthKit, or both to Next.js applications | | Jest operations | 12 | Inspecting, running, configuring, and troubleshooting common Jest operations | | LLM costs | 3 | Single-thread, project-scoped, and machine-wide LLM cost, effort, and timing reports | | MongoDB operations | 10 | Common MongoDB Shell and Database Tools work with read-first safety checks | | Node.js API operations | 10 | Inspecting, changing, testing, and troubleshooting Node.js APIs across common frameworks | | Super | 8 | Repeatedly completing general or specialist project tasks to a high evidence-based bar | | Security | 1 | Read-only dependency vulnerability and supply-chain auditing with critical host-harm triage | | Utilities | 8 | Code-quality review and improvement, Markdown compression, reverse engineering, critique, analytics, publication CMS, calls to action, and project sites | | Web design | 3 | The 11ai visual language and deliberate content-led styling for accessible web interfaces | | Cross-harness | 2 | Delegating work to agents running in other CLI harnesses |

Start with a plugin's README to choose a skill, then name that skill in your request. Skills provide instructions and examples; they do not install the application dependencies used in those examples by themselves.

Repository layout

.claude-plugin/               Claude marketplace discovery entry point
v0/
  index.js                    CommonJS package entry point
  plugins/
    11ai-agent-automation/    8 automation skills under skills/
    11ai-ai-chat/             9 AI chat skills under skills/
    11ai-antdesign-operations/ 10 Ant Design React operation skills under skills/
    11ai-aws-operations/      11 AWS CLI operation skills under skills/
    11ai-benchmarks/          19 benchmark skills under skills/
    11ai-blog-builder/        8 editorial blog skills under skills/
    11ai-cleanup/             5 cleanup skills under skills/
    11ai-codebase/            3 codebase skills under skills/
    11ai-docker-operations/   10 Docker CLI operation skills under skills/
    11ai-git-operations/      12 Git operation and workflow skills under skills/
    11ai-integrations/        3 integration skills under skills/
    11ai-jest-operations/     12 Jest operation skills under skills/
    11ai-llm-costs/           3 LLM cost, effort, and timing skills under skills/
    11ai-mongodb-operations/  10 MongoDB operation skills under skills/
    11ai-nodejs-api-operations/ 10 Node.js API operation skills under skills/
    11ai-super/               8 iterative task and improvement skills under skills/
    11ai-security/            1 dependency security audit skill under skills/
    11ai-utils/               8 utility skills under skills/
    11ai-web-design/          3 web design skills under skills/
    11ai-xharness/            2 cross-harness skills under skills/
  scripts/                    Package validation, release, and publishing helpers
  www/                        Next.js project site

v0 is the current versioned distribution namespace for the plugins, package tooling, and project site. The root marketplace manifest is their stable Claude discovery entry point. Pin the npm package version or a commit when reproducibility matters, because its content can change between releases. See CHANGELOG.md for release history.

The CommonJS entry point exposes package metadata only:

const elevenAI = require("@rj11io/11ai")

console.log(elevenAI.name) // "@rj11io/11ai"

The supported consumer surface is the skill content under v0/plugins/{plugin-name}/skills/{skill-name}, not a JavaScript runtime API.

Repository commands

Install root package tooling before running these commands:

npm install
npm run validate-skills
npm run pack-dry

validate-skills checks frontmatter, Codex metadata, Claude plugin and marketplace configuration, links, scripts, and catalog coverage. pack-dry shows the npm tarball contents without publishing. The manual publishing command and its token requirements are documented in v0/scripts/README.md. The project site has its own dependencies and commands in v0/www/README.md.

Pushes to main run semantic-release, which updates the changelog and package version, publishes to npm, creates a GitHub release, and publishes the generated tarball to GitHub Packages.

License

Apache-2.0. See LICENSE.