Podium · Podium

The documentation home. One catalog, every harness, and where each feature, tier, and guide is documented.

Podium holds reusable AI agent artifacts in one catalog and translates them into the file layout each runtime expects. An author writes an artifact once in the canonical format, and a harness adapter produces the Claude Code layout, the Cursor layout, or a published marketplace repository from it.

An artifact is a directory, and the directory carries its dependencies. A script bundled beside a SKILL.md reaches every layout the manifest reaches.

This page is the documentation home. It names the pages that cover each feature, each deployment tier, and each task.

Note

Status: 0.3.x, early release. The CLI, server, MCP bridge, and SDKs are published. Install with brew tap lennylabs/tap && brew install podium (macOS / Linux) or scoop bucket add lennylabs https://github.com/lennylabs/scoop-bucket && scoop install podium (Windows). See Implementation status for what's shipped and what's still on the roadmap to 1.0.

Start here

Page	What it covers
Why Podium	What "one catalog, every harness" means feature by feature, when Podium applies, when a simpler alternative is enough, and how it compares to adjacent products.
Quickstart	Install the CLI, write one skill, materialize it into Claude Code, and see it load.
Concepts	The vocabulary the rest of the docs link to: artifact, type, domain, registry, layer, visibility, harness, profile, materialization, and meta-tools.
How it works	The components, where each feature runs, the deployment tiers, and where state lives.

Features

Feature	What it does	Covered in
Cross-harness delivery	A harness adapter translates a canonical artifact into the layout one runtime expects, for a workspace tree or a published marketplace repository.	Configure your harness, Marketplace publishing
Domains and subdomains	The directory layout defines the domain hierarchy, and `DOMAIN.md` adds descriptions, keywords, and featured artifacts.	Domains
Selective materialization	`podium sync` materializes the subset named by include globs, exclude globs, and types, and a named profile replays that subset.	Selective materialization
Progressive discovery	An agent traverses domains and searches the catalog through the meta-tools, and materializes an artifact and its bundled files only when it loads one.	Browsing the catalog
Layered composition	Several independent sources compose into one catalog with deterministic merge, explicit precedence, and `extends:`.	Layered composition
Access control	Each layer declares who can see it, and the registry composes the caller's effective view from the layers that pass.	Access control

Deployment tiers

Each tier keeps everything the tier below it does and adds server-side capability. The artifacts are the same in every tier.

Tier	Server-side deployment	What the tier adds
Local	None	Authoring, lint, sync, domains, profiles, and ordered layers from disk
Single node	One binary	Discovery through MCP or the SDKs, hybrid search, registered and remote layers with visibility, and one audit log
Clustered	Replicas, Postgres, and object storage	Multi-tenancy, SCIM group sync, signing with a transparency log, and high availability

Server-side integrations names the backing service behind each server-side concern: the metadata store, object storage, the vector index, embeddings, identity, and layer sources. Each row states what ships by default and what can replace it.

Guides

Guide	What it covers
Getting Started	Positioning, the quickstart, the vocabulary, and the architecture.
Authoring	Writing artifacts: types, frontmatter, domains, bundled resources, `extends:`, rule modes, hooks, and hints.
Consuming	Delivering artifacts into a harness, narrowing what a workspace receives, runtime discovery, the SDKs, and marketplace publishing.
Deployment	Running each tier, the server-side integrations, layers, access control, day-two operations, and the OIDC cookbooks.
Reference	CLI, HTTP API, frontmatter schema, error codes, and glossary.
Testing	Setting up and running the integration and live-backend suites.
About	Implementation status, contributing, governance, and the changelog.

'Hello world' example

The full quickstart covers the same flow with prerequisites and verification steps.

Create a skill directory with a SKILL.md file for agent-facing instructions and an ARTIFACT.md file for Podium metadata:

~/podium-artifacts/personal/hello/greet/SKILL.md

---
name: greet
description: Greet the user by name and tell them today's date.
---

Greet the user by their first name. Tell them today's date.

~/podium-artifacts/personal/hello/greet/ARTIFACT.md

---
type: skill
version: 1.0.0
tags: [demo, hello-world]
---

<!-- Skill body lives in SKILL.md. -->

Point Podium at the directory and set the harness:

shell

cd ~/projects/your-project
podium init --registry ~/podium-artifacts/ --harness claude-code
podium sync

Open Claude Code in the project. Claude Code can discover the materialized skill in its native location. Changing the harness and running podium sync again writes the same artifact into another runtime's layout, and it removes the files the previous harness's run wrote, because each run reconciles the whole target against the lock file.

The source, the issue tracker, and the release history live at lennylabs/podium.