Atomic ships 57 skills underDocumentation Index
Fetch the complete documentation index at: https://bastani.mintlify.app/llms.txt
Use this file to discover all available pages before exploring further.
.agents/skills/<name>/SKILL.md. They auto-invoke when the agent detects a relevant trigger, or you can call them directly with /<skill-name> (Claude Code) or natural language (OpenCode / Copilot CLI).
Skills are auto-discovered by every supported agent. All three agents share the same skill set via .agents/skills/ — see agent-specific files.
Top skills to know first
| Skill | Invoke with | Purpose |
|---|---|---|
init | /init | Generate CLAUDE.md / AGENTS.md by exploring the codebase. |
prompt-engineer | natural language | Sharpen your research prompts, workflow inputs, or any agent prompt before you run it. |
research-codebase | /research-codebase "<question>" | Dispatch parallel sub-agents to analyze the codebase and write a research doc. |
create-spec | /create-spec "<research-path>" | Produce a technical execution spec grounded in a research document. |
workflow-creator | natural language | Generate a multi-agent workflow definition using defineWorkflow() + a registry. |
tdd | natural language | Red-green-refactor with a built-in testing-anti-patterns guide. |
explain-code | /explain-code "<path>" | Deep-dive explanation of specific code using DeepWiki. |
gh-create-pr | /gh-create-pr | Commit, push, and open a GitHub PR (also /ado-create-pr, /sl-submit-diff). |
playwright-cli | natural language | Automate browser interactions, tests, screenshots. |
impeccable | natural language | Create distinctive, production-grade frontend interfaces. |
find-skills | natural language | Discover and install community skills you don’t have yet. |
Full catalog
All 57 skills, grouped by category:Development workflows
Development workflows
init, research-codebase, create-spec, workflow-creator, explain-code, find-skills, tdd, prompt-engineerContext engineering
Context engineering
context-fundamentals, context-degradation, context-compression, context-optimization, filesystem-context, memory-systems, multi-agent-patterns, tool-design, hosted-agents, project-development, bdi-mental-statesTypeScript & runtime
TypeScript & runtime
typescript-expert, typescript-advanced-types, typescript-react-reviewer, bun, opentuiFrontend design & UI polish
Frontend design & UI polish
impeccable, polish, critique, audit, layout, typeset, colorize, adapt, animate, delight, clarify, distill, quieter, bolder, overdrive, harden, optimize, arrange, extract, normalize, onboard, shape, teach-impeccable, frontend-design, ux-design-virtuosoEvaluation
Evaluation
evaluation, advanced-evaluationDocuments & parsing
Documents & parsing
pdf, xlsx, docx, pptx, liteparseSource control & automation
Source control & automation
gh-commit, gh-create-pr, ado-commit, ado-create-pr, sl-commit, sl-submit-diff, playwright-cliMeta
Meta
skill-creatorls .agents/skills/ for the live, on-disk list.
Source-control MCP servers are disabled by default. Set
scm in .atomic/settings.json (or run atomic config set scm <provider>) to github, azure-devops, or sapling to enable the matching MCP server. sapling disables both.Next
Sub-agents
The 12 specialized sub-agents that skills and workflows dispatch under the hood.
Context files
How
/init and the research → spec flow generate CLAUDE.md / AGENTS.md.