AgentGem documentation

AgentGem is a local web UI that introspects your coding-agent config — skills, MCP servers, and CLAUDE.md — redacts secrets at capture, and builds a secret-safe, composable Gem. A Gem is a portable archive (manifest + lock) you can publish to a GitHub-backed registry, merge with others, and deploy to several targets.

It's built on AgentBack, the ninemind AI-native API/MCP framework: every operation is defined once as a Zod contract and exposed as a REST endpoint, an MCP tool, and an OpenAPI 3.1 document — so the web page and your local agent call exactly the same thing.

Start here#

• Getting started — install, run the local server, and build your first Gem.
• Concepts — what a Gem is, the archive format, the redaction trust boundary, and the AgentBack one-contract model.

Architecture & internals#

• Architecture — the system map: clients → contract surface → Gem core → distribution, with diagrams.
• The build pipeline — introspect → redact → buildGem → archive.
• Archive format — the manifest + lock spec, hashing, and serialization.
• Redaction — the trust boundary and its rules.
• API reference — every REST endpoint and MCP tool.

Distribution#

• Targets & deploy — the deploy targets (Eve, Flue, OpenAI Sandbox, Bedrock AgentCore) and the publish / undeploy lifecycle.
• Registry — the GitHub-backed Gem registry: publish, resolve, merge, and install composable Gems.
• Testbed & run — install a Gem into a local testbed; run or deploy a materialized target locally, to Vercel, or to Cloudflare.

Contributing#

• Development — build, test, the decorator/compiled-dist setup, and code conventions.

Diagrams#

Architecture diagrams live in diagrams/ as .svg (embedded in these docs), .png (raster fallback), and .html (interactive, with Copy / PNG / PDF export):

• System architecture
• Gem build pipeline
• Distribution