Codebelt

Agentic skills for disciplined
AI-era engineering

Codebelt skills are modular, expert-level capabilities designed for AI agents. They encapsulate best practices, engineering principles, and automated workflows into reusable units that empower agents to perform complex software development tasks with precision.

Friendly robot wizard casting glowing code magic for AI agent skills. Friendly robot wizard casting glowing code magic for AI agent skills.
56 Total installs
14 Available skills
4 Skill groups
Universal Compatibility
.NET · NuGet · Engineering

.NET Change Impact

A release-classification skill that turns compatibility analysis into a structured SemVer recommendation.

.NET · NuGet · Engineering

.NET DocFX Digest Steward

A documentation-maintenance skill for teams that want DocFX pages to stay useful after the API changes.

.NET · NuGet · Engineering

.NET Application Solution Setup

A solution-scaffolding skill for new .NET applications that need conventions, not just generated files.

.NET · NuGet · Engineering

.NET Library Solution Setup

A library-scaffolding skill for maintainers who want a NuGet repository that is ready for stewardship, not just compilation.

.NET · NuGet · Engineering

Strong Name Signing for .NET

A focused skill for creating `.snk` files and wiring assembly signing into modern .NET workflows.

Git · Releases · Automation

Git Keep A Changelog

A changelog-authoring skill that turns raw git history into a curated release narrative.

Git · Releases · Automation

Git NuGet README

A README-writing skill that helps a package explain itself honestly and convincingly.

Git · Releases · Automation

Git NuGet Release Notes

A package-release-note skill for teams that need per-package history, not just a repo-wide changelog.

Git · Releases · Automation

Git Remote Release

A release-note skill for turning compare ranges into polished GitHub-ready summaries.

Git · Releases · Automation

Git Repo Digest

A digest-generation skill for maintainers who need grounded narrative from a real codebase, not generic repository prose.

Git · Releases · Automation

Git Visual Commits

A commit orchestration skill for teams that want commit quality and attribution treated as part of the workflow.

Git · Releases · Automation

Git Visual Squash Summary

A read-only summarization skill for preparing squash-and-merge history without committing anything.

Content · Documentation

Markdown Illustrator

A content-to-visual skill for teams that want markdown transformed into a clear visual concept instead of another summary paragraph.

Repository · Workflow

Trunk-First Repo

A repository-setup skill for teams that want healthy git habits and branch policy from the first commit onward.

Four groups, one expertise.

.NET · NuGet · Engineering

Expert-level .NET engineering at your fingertips. From Library Scaffolding and App Scaffolding to DocFX Digests and Strong-Name Signing. Built to enforce Codebelt standards across your entire solution.

Git · Releases · Automation

Automate your software lifecycle. From Changelog Automation to Remote Releases and Visual Commits. Hardened workflows for GitHub and NuGet, driven by MinVer and Git tags.

Content · Documentation

Turn raw information into polished artifacts. Markdown Illustrator transforms documents into visual briefs and diffuser prompts for rich communication.

Repository · Workflow

Establish healthy source-control habits from day one. Trunk-First Repository sets up scaled trunk-based development with PR-first workflows and branch protection.

The Codebelt experience

Why consistency matters in AI-assisted engineering.

Codebelt skills are built for AI-assisted delivery where the model can change but the workflow should still feel familiar. Frontier and open-weight model families, terminal agents and editor agents all benefit from the same opinionated structure: source-grounded instructions, explicit guardrails, and repeatable execution that reduce drift between runs.

Consistency is keyShared workflows, naming, and release discipline keep outcomes recognizable from one skill to the next.
Frontier + open-weight readyAuthored to travel cleanly across frontier and open-weight models, so changing the reasoning engine does not mean rebuilding the process.
Source-grounded by defaultSkills are designed to inspect code, docs, Git history, and live metadata before they generate conclusions or edits.
Tool-native executionBuilt for real terminal and IDE agents that can search, edit, build, test, and release instead of stopping at advice.
Opinionated guardrailsVerification steps, repository conventions, and release-aware wording help reduce hallucinated shortcuts and accidental drift.
End-to-end coverageFrom scaffolding and documentation to changelogs, release notes, and repository hygiene, the belt keeps the SDLC connected.
Agentic OpenAI Anthropic Google Z AI DeepSeek Xiaomi MiniMax Deterministic