2lab-aiOfficial·15 Agent Skills Included

oh-my-claude

Autonomous multi-agent coding loops with spec-driven verification

Runs autonomous development loops that plan, implement, and verify code changes until tasks are genuinely complete. Turns vague feature requests into structured specs, traced scenarios, and contract tests before any code is written. Eliminates manual review cycles with triple-model scoring gates and cross-tool session saving. Coordinates Gemini, Codex, and Claude subscriptions so work continues seamlessly across different coding environments.

36|3

↗

npx skills add 2lab-ai/oh-my-claude --all -g -y

All Skills in This Repository (15)

Pure Emerald Level Indicators

📦 In Repo

2lab-ai

inductive-distillation

Distill real-world experiences into compact, repeatable process templates.

do-work

Automates unfinished STV trace scenarios from docs/*/trace.md through stv:work and quality gates.

trace

Generate vertical traces and RED contract tests from feature specs.

using-terminal-charts

Render numeric data as terminal charts using chartli.

excalidraw-diagram

Generate Excalidraw JSON diagrams from natural language descriptions.

new-task

Decompose vague requirements into STV-based specs and trace artifacts.

stv-verify

Compare issue acceptance criteria against GitHub PR changes to report gaps.

clarify

Resolve ambiguous user requests into a concise Context Brief via Q&A and codebase exploration.

what-to-work

Scan docs/*/trace.md for unfinished scenarios and route to next-step skills.

spec

Generate a formal feature spec with decision gates and changelog.

explore

Explore codebases read-only to map architecture, dependencies, and knowledge gaps.

plan-new-task

Propose development tasks and generate STV specs and trace artifacts from git history.

Official

Intermediate

Frequently Asked Questions

FAQPage Schema

How to install oh-my-claude?▼

Run `npx skills add 2lab-ai/oh-my-claude --all -g -y` in your terminal to install all skills in this suite globally.

What does oh-my-claude actually do?▼

It runs autonomous work loops where multiple models plan, code, and review changes until a task is verifiably complete. Commands like /ultrawork and /deepwork iterate until quality gates pass.

How does the STV workflow prevent bad code?▼

Every feature goes through spec, trace, work, and verify phases with contract tests and gap detection. Code must match the written trace exactly before it is marked complete.

Can I move work between Claude Code and other tools?▼

Yes. The /save and /load commands store work context in a shared folder so you can resume the same task in Gemini CLI or Codex.

Do I need coding experience to use oh-my-claude?▼

Basic familiarity helps, but the skills guide you through interviews and make autonomous decisions on small choices. You only get asked about decisions with significant switching costs.