Built into the lpm desktop app
lpm lists the skills Claude Code and Codex load in each project, shows what each one costs in context, and gives you one dialog to create and edit them, with AI drafting the fields from a plain-English description.
A skill is a SKILL.md file in a dotfolder — easy for agents to read, fiddly for you to maintain. lpm turns it into a form, with an AI Draft bar that fills every field from one sentence.
Saved straight into a folder the CLI you pick already reads — no wondering where the skill goes.
Your agent, when the description fits, or manual-only, invoked with /name or $name.
The only part your agent reads before deciding to open the skill. lpm shows its per-turn token cost.
Read only once the skill runs, so their length costs nothing up front.
Claude Code and Codex each read skills from their own folders. lpm lists the skills in each one, plugin skills included, and writes new skills where the CLI you pick will read them.
lpm writes a standard SKILL.md, so everything works exactly as if you had written it by hand.
The same dialog writes Codex skills, including the shared folder other agent CLIs read too.
Type one sentence about the task. lpm reads your repository and drafts the name, description, and instructions to match how your project works, on the AI CLI and model you pick. Nothing is saved until you click Create.
Reopen any skill in your personal or project folders, including ones you wrote by hand, and change its description, instructions, or who runs it. lpm rewrites only the fields you touched.
Descriptions load before every turn; instructions only when a skill runs. lpm estimates what each description costs every turn and shows each skill's size on disk, so you know which deserve a manual-only switch.
Press ⌘⇧K (Ctrl+Alt+Shift+K on Windows and Linux) in any project, or choose Skills & tools from the ⋯ menu in the terminal pane header. You can pin it to the toolbar from there.
Click New skill, describe the task, and let AI draft the fields, or fill in the name, description, and instructions yourself.
Pick who runs it and press Create. Reopen the skill any time to edit it, and a deleted skill goes to the Trash (the Recycle Bin on Windows), so you can get it back.
The pane that creates skills also shows everything else Claude Code and Codex will load in that folder, and what it costs.
Skills sit next to subagents, slash commands, MCP servers, plugins, hooks, and CLAUDE.md or AGENTS.md files, for Claude Code and Codex, filterable by CLI, kind, or text.
A running estimate of what the agent loads every turn, from instruction files and skill and subagent descriptions. MCP tool schemas are flagged as not counted.
lpm warns when a name is already taken in that folder, or when your personal copy of a skill will win over the project's copy.
Skills that ship inside a Claude Code plugin are listed and readable but not editable, and on SSH projects the whole pane is a read-only view.
Before a skill is removed, a preview shows exactly what goes to the Trash (the Recycle Bin on Windows).
One click in Settings installs skills that teach Claude Code, Codex, Gemini CLI, and OpenCode to start, stop, and duplicate your lpm projects.
How skills work, what they cost in context, and what lpm writes to disk.
Pick presets, reorder fields, tune colors and meters, and watch the statusline update live.
Pin an account to each project; skills, subagents and commands are shared across all of them.
Tokens and cost by project, plus live 5-hour and weekly limit meters, in a private desktop app.
Give Claude Code and Codex tools to run services, inspect logs, and work across project copies.
Run multiple AI coding agents while every project, service, and terminal stays visible.
Every panel in the app, from services and review to automations and remote machines, on one page.
Skills, services, logs, Git changes and project copies for Claude Code and Codex, all inside one native desktop app.