
Skills
Open-source, composable agent skills for real software engineering across planning, TDD, domain modeling, triage, and code review. Install via skills.sh or a Claude Code plugin, integrate with GitHub or Linear, and standardize feedback loops for reliable, production-quality changes.
Overview
A typical session begins with grill-with-docs to align intent and sharpen terminology, updates CONTEXT.md and ADRs, then proceeds with tdd at pre-agreed seams. Work is split via to-tickets, reviewed with code-review, and merged after implement closes the loop against the originating issue.
What It Does
Ideal for senior and staff engineers, tech leads, maintainers, and platform teams who want agents to behave predictably in established repositories. It also suits founders and agencies pressing for reliable delivery across multiple clients, and individual developers adopting disciplined practices. Use it when you need rigorous specs, safer refactors, high-signal reviews, and reproducible debugging, or when a codebase risks collapsing into a ball of mud.
- Runs a structured grilling interview to eliminate ambiguity before any code changes.
- Builds and maintains a shared domain glossary and ADRs alongside evolving decisions.
- Enforces red-green-refactor discipline with tight loops and pragmatic test heuristics.
- Generates tracer-bullet tickets, dependencies, and wayfinding maps across large projects.
- Audits architecture, proposes deep modules, and guides incremental refactors without churn.

Where It Helps
Editor’s View
Getting started takes minutes. Add the skills to a repository with the skills.sh installer, then run the setup skill once to select your issue tracker (GitHub, Linear, or local files), declare triage labels, and choose where docs should live. Prefer a managed path? Install the Claude Code plugin variant to subscribe to an auto-updating bundle. From there, begin with grill-with-docs, drive changes with tdd at pre-agreed seams, split work using to-tickets, and close with code-review and implement. The same flows apply across agents that support the Agent-Skills standard.
Disciplined, model-agnostic skills that make agents behave like great teammates.
Getting Started
Unlike monolithic agent frameworks, Skills stays small, composable, and auditable. Each capability is a focused discipline you can run, remix, or replace, and every artifact lives alongside your code and tickets. Tight loops, explicit trackers, and architecture-first practices compound over time, improving throughput without sacrificing correctness. If you want agents that ship trustworthy changes instead of verbose drafts, this is a pragmatic, production-ready path.
Open the tool and review its core product experience.
Create your account or access your existing workspace.
Use your own task to judge speed, quality, and fit.
Check similar AI tools before making a final decision.


Comments (0)
No Comments Found