Context Engineering Kit
A plugin marketplace for agents: reflexion, code and PR review, git, TDD, DDD, specs and subagent-driven development
Medium risk
We rate an entry medium when the tool runs code, makes network calls or reads project files. Check what exactly it does before installing.
Why this level
- Plugins read and change project code and run sub-agents
- Git commands create commits and pull requests in repositories
Install
Manual install
/plugin marketplace add NeoLabHQ/context-engineering-kit
/plugin install reflexion@NeoLabHQ/context-engineering-kitInstall plugins by name; each loads only its own agents, commands and skills.
This is third-party code. Review the repository files before installing.
What it does
The set gives an agent a marketplace of plugins with context engineering techniques so results are more predictable and higher quality. Plugins install separately and do not overlap: reflexion adds self-check and refinement loops over a response, review brings code and pull request review through several specialized sub-agents, git adds commit and PR commands plus skills for worktrees and notes, tdd and ddd add testing and domain-driven design practices with rules, sdd and sadd drive spec-based and subagent-based development with review between tasks, and kaizen adds root-cause analysis commands. Each plugin loads only its own agents, commands and skills. There is a ready CI action and optional external MCP setup for tools such as arxiv and context7 through separate setup skills.
Who it is for. For developers and tech leads who work with coding agents and want steady output quality.
Good fit when
- You want the agent to re-check and refine its own answer
- You want code or pull request review from several sub-agents
- You want spec-driven or subagent-driven development with review between tasks
Not a fit when
- The task is not about development or coding agents
- You want minimal context rather than a set of process plugins
Example request
Implement user authentication, then reflect and fix the issues you findLimitations
The full feature set targets Claude Code with sub-agent, command and hook support. In Gemini CLI and Antigravity the plugins install as one bundle without per-plugin selection, and installation via npx skills does not support sub-agents, so the experience is partial. Some skills connect external MCP servers and need those tools. Material is in English. GPL-3.0 licensed.
How to disable. Remove the installed plugins via /plugin, and if needed drop the marketplace with /plugin marketplace remove context-engineering-kit.
Security check
- Plugins read and change project code and run sub-agents
- Git commands create commits and pull requests in repositories
README in short
The README presents a marketplace of context engineering plugins for Claude Code, OpenCode, Cursor, Antigravity and other agents. It states its principles: simple install with no dependencies, token efficiency, a focus on result quality, per-plugin granular install and reliance on proven techniques and the agentskills.io standard. The set includes the reflexion, review, git, tdd, sadd, ddd, sdd, kaizen and other plugins. Installation is shown for Claude Code via the plugin marketplace, for Gemini CLI and Antigravity as a single bundle, and for others via npx skills and openskills. It gives an example using the /reflect and /memorize commands. GPL-3.0 licensed.
FAQ
Do I have to install every plugin?
No. Each plugin installs separately and loads only its own agents, commands and skills into context, without overlap.
Does it work outside Claude Code?
There is installation for Gemini CLI, Antigravity, Cursor, Codex and others via npx skills, but there the plugins install as a bundle and without sub-agents, so some features are unavailable.
Related
Open-source personal AI assistant on your own machine: answers in Telegram, Slack, Discord and WhatsApp, extended with skills and plugins
A self-improving agent from Nous Research with a TUI, messaging gateway, cron jobs and skills it writes itself
An open source coding agent for the terminal and desktop with build and plan modes
Open prompt library with a Claude Code plugin, MCP server and CLI: search, fetch and improve prompts and skills from an agent