ravid7000/skills

Agent skills for real engineering work.

Small, sharp instructions your coding agent follows — install the ones you want, then get back to shipping.

skills.sh install badge for ravid7000/skills

Install

Pick the skills you use. The installer writes them into your agent.

npx skills add ravid7000/skills

View on GitHub skills.sh

Claude Code · Cursor · Codex · Copilot · MIT

01

Skills.

Grouped by when you reach for them. Most people start with planning.

01 Workflow

Plan the work, then hand it off cleanly when a session ends.

Start with /plan-with-me

02 Research

Answers from current sources, not model memory.

Start with /finder

03 Diagnostics

See what’s broken — in a single repro or in production.

Start with /debug-flow

04 Meta

Author skills that load reliably and survive review.

Start with /creating-agent-skills

02

What is a skill?

The problem

An agent is only as good as the process you give it. Left to guess, it writes plausible code that quietly drifts from your standards.

The fix

A skill encodes one good habit — planning, researching, tracing a bug — so the agent runs it the same way every time.

Why it compounds

Skills are plain files, not a lock-in platform. Tune one step and the whole workflow gets sharper.

Works in whatever agent you already use · Claude Code · Cursor · Codex · Copilot · MIT

03

What changed.

Full changelog
v0.4.1 debugging-ui-flows becomes debug-flow

Renames the diagnostics skill to debug-flow and generalizes it beyond UI→API journeys to any broken flow across a stack.
v0.4.0 handing-off-work

A session-end handoff a fresh agent can resume from — decisions, dead ends, and verified code state.
v0.3.0 plan-with-me and debugging-ui-flows

Interactive planning that settles decisions one question at a time, plus temporary UI→API flow tracing for hard bugs.

Install the skills
and get to work.

Everything is MIT licensed and lives in one repo.

npx skills add ravid7000/skills

View on GitHub skills.sh

Claude Code · Cursor · Codex · Copilot · MIT