agentic-awesome-kits
Health Warn
- License — License: NOASSERTION
- Description — Repository has a description
- Active repo — Last push 0 days ago
- Low visibility — Only 6 GitHub stars
Code Pass
- Code scan — Scanned 12 files during light audit, no dangerous patterns found
Permissions Pass
- Permissions — No dangerous permissions requested
No AI report is available for this listing yet.
A universal collection of reusable Loop Engineering kits for AI coding agents—not just prompts. Each kit combines Skills, Rules, Templates, workflows, and validation loops to help agents autonomously plan, execute, verify, fix, and improve engineering tasks.
Agentic Kits
Reusable AI Skill Kits for coding agents (Claude Code and any other subagent/slash-command-compatible tool). Each kit is a self-contained package of Skills (slash commands), Rules (conventions the agent must follow), and Templates (fill-in-the-blank output baselines) — copy .claude/ into your project and the skills work immediately.
/dockerfile .NET 10 Minimal API --prod
/deploy container-api --namespace prod --replicas 3 --with-ingress
/log-check booking-api --env prod --since 1h
Note — examples use .NET 10.
Sample commands,Dockerfile.dotnet, and related container templates in this repo target .NET 10 (e.g. Minimal API). Skills are stack-agnostic: pass another runtime in the prompt when you need it.
Browse the visual index: index.html · DevOps docs: devops/
Live site (GitHub Pages + Jekyll):
https://KhaiTrang1995.github.io/agentic-kits/
Why this exists
Prompting an agent to "write me a Dockerfile" gets you a Dockerfile — not necessarily a multi-stage, non-root, healthcheck'd one your team would actually ship. A kit turns that tribal knowledge into a rules/*.md file the agent reads before generating anything, so the output is consistent whether it's you, a teammate, or a different agent session asking.
Categories
| Category | Status | Kits |
|---|---|---|
| devops | ✅ Available | Run: docker-kit, docker-swarm-kit, k8s-kit · Observe: zabbix-kit, elk-kit (/log-check), grafana-kit (/loki-query) |
| software | ✅ Partial | drawio-kit — .drawio architecture / flowchart / ERD / advanced diagram · more role kits TBD |
DevOps at a glance
| Kit | Skills (highlight) |
|---|---|
| docker-kit | /dockerfile, /compose, /ci |
| docker-swarm-kit | /stack, /service, /secret |
| k8s-kit | /deploy, /helm, /monitor |
| zabbix-kit | /zabbix-check, /zabbix-trigger, … + ZABBIX-BRAIN.md |
| elk-kit | /log-check, /es-query, /logstash, … + ELK-BRAIN.md |
| grafana-kit | /loki-query, /grafana-dashboard, /grafana-alert, … + GRAFANA-BRAIN.md |
Log check quick map
Elasticsearch / Kibana → elk-kit /log-check
Loki / Grafana → grafana-kit /loki-query
Zabbix problem / log item → zabbix-kit /zabbix-check
Software at a glance
| Kit | Skills (highlight) |
|---|---|
| drawio-kit | /architecture, /flowchart, /erd, /diagram (draw.io CLI for advanced) |
Quick start
git clone <this-repo>
cp -r agentic-kits/devops/docker-kit/.claude/ your-project/.claude/
cp -r agentic-kits/devops/docker-kit/_templates/ your-project/_templates/
# In Claude Code (or any compatible agent) — example uses .NET 10:
/dockerfile .NET 10 Minimal API --prod
You can copy multiple kits into the same project. Watch for skill-name collisions across kits (rare, but check .claude/skills/ before merging).
Structure of every kit
{kit-name}/
.claude/
skills/
{skill-name}/
SKILL.md ← trigger, input examples, process, output, references
rules/
{convention}.md ← what the agent must/must never do when generating output
_templates/
{output-template} ← ready-to-fill baseline (Dockerfile, .yml, ...)
references/ ← optional cheatsheets (observability kits)
*-BRAIN.md ← optional long-lived guidance (observability kits)
README.md ← how to use this specific kit
Design principles shared across every kit
- The agent never runs the risky command. It generates
Dockerfiles, manifests, queries, and checklists — it does not rundocker build,kubectl apply,helm install, or mutate prod monitoring systems. A human reviews and runs those. See each kit'sapproval-gate.md. - Rules are the single source of truth. Skills reference a shared
*-conventions.mdfile instead of repeating the same security/resource rules inside every skill. - Approval gate, not a wall of Y/n. Every kit follows the same L1 (plan preview) → L2 (diff on edit) → L3 (iterate on creative output) pattern, so switching between kits doesn't mean relearning a new interaction model.
- Kits compose. Where it makes sense, one kit references another instead of duplicating content — e.g.
docker-swarm-kitpoints atdocker-kit's Dockerfile conventions rather than restating them. - Evidence-only on investigations. Log and problem checks do not invent hit counts or metric values.
Contributing
Want to improve a kit, add a skill, or fix docs? See CONTRIBUTING.md for structure, checklists, and PR guidelines.
GitHub Pages (Jekyll)
This repo publishes a static docs site with Jekyll (GitHub Pages).
| File | Role |
|---|---|
_config.yml |
Site title, baseurl, exclude rules for kit internals |
Gemfile |
github-pages + plugins for local parity |
.github/workflows/pages.yml |
Build & deploy on push to main |
index.html / devops/index.html |
Docs UI |
Enable once: repo Settings → Pages → Build and deployment → Source: GitHub Actions.
Local preview
# Ruby 3+ recommended
bundle install
bundle exec jekyll serve
# open http://127.0.0.1:4000/agentic-kits/
Kit skill packages (
.claude/, templates, brains) are excluded from the Jekyll site so they stay source-only on GitHub. Use the HTML docs for browsing; clone the repo for skills.
License
CC BY-NC-SA 4.0 — share, adapt, and build on these kits freely, including commercially-adjacent internal use, but do not resell them and keep derivatives under the same license with attribution.
Roadmap
- Observability kits — Zabbix, ELK (
/log-check), Grafana/Loki (/loki-query) -
software/drawio-kit— technical diagrams as.drawio - More
software/kits — BA, PM, dev (per-stack), QA, tech lead - Cross-kit index / search once the kit count grows past a handful per category
Reviews (0)
Sign in to leave a review.
Leave a reviewNo results found