Organization Skill Management
Create and edit skills in the Registry UI or API, organize them with categories and tags, track versions, and attach supporting text or binary files.
Enterprise Skill Gateway for Governed AI Skills
Privately host organization-curated AI skills, govern who can access them, and distribute approved content consistently to the developer tools your teams already use. One Registry control plane for authoring, importing, managing, and sharing skills without publishing private content publicly.
Components
What's Inside the Skill Gateway
Four capabilities cover how approved skills are authored, governed, imported, and distributed across your organization.
Skill Access & Governance
Skills are private by default. Resource-level ACL permissions control who can view, edit, delete, or share each skill before content is delivered.
Provider Sync
Import skills one-way from your Git provider. Discover SKILL.md files, extract an immutable snapshot, apply valid changes atomically, and report errors per skill.
Private Developer Distribution
Deliver approved skills to personal or project environments through the AI Skills CLI, with the Skill Gateway as the source of truth.
Features
Enterprise-Grade Skill Gateway Capabilities
Everything platform teams need to author, govern, import, and deliver AI skills from one private Registry.
01 — Skill Authoring
Manage Skills in the Registry
Create and edit skills through the Registry UI or API. Organize them with categories and tags, track a version, and enable or disable a skill as its availability changes. Each skill can include its SKILL.md content and supporting text or binary files.
Supporting files are validated before they are stored, which keeps skill content predictable for both the Registry and downstream sync clients.
incident-triagecategory: operations
v1.4.0Enabled
- SKILL.md4 KB
- references/runbook.md18 KB
- scripts/triage.py9 KB
- assets/escalation.png312 KB
5 MiBper file50 filesper skill10 MiBtotal payload
02 — Access Governance
Govern Access with Resource-Level ACLs
Skills are private resources by default. The skill creator receives full permissions, while other users receive only the access explicitly granted to them. Listing and content endpoints filter by ACL, so an inaccessible skill is never presented as an available delivery target.
Explore Governed AI Layer03 — Provider Sync
Import Skills from Any Git Provider, One Way
Connect your Git provider through an app or OAuth connection so the Registry can read a repository and import skills in one direction: repository to Registry. The pipeline resolves the selected ref to a commit, downloads an immutable snapshot, and discovers skill folders containing SKILL.md.
Extraction is protected against path traversal, links, oversized files, and decompression bombs. Each skill is validated before the batch is applied atomically, and item-level errors are retained in the job report.
- Git reporef → commit
- Snapshotimmutable copy
- DiscoverSKILL.md folders
- Validateparse per skill
- Jarvis RegistrySkill Gatewayatomic apply · per-skill errors
One-way import · the Registry never pushes changes back to the provider
04 — Developer Distribution
Deliver Approved Skills to Developer Tools
The AI Skills CLI is the delivery layer for accessible Registry skills. It syncs approved content into personal or project environments for Claude Code, Codex, and GitHub Copilot, so every developer works from the same governed source.
Share approved content with developers without publishing private skills to a public marketplace.
Explore AI Skills CLI
- Skill Gatewaysource of truthACL-filtered
- AI Skills CLIsync
- Developer tools
- Claude Code
- Codex
- GitHub Copilot
Synced into personal or project environments
At a glance
Skill Gateway
The Skill Gateway is the private control plane for organization-curated AI skills. These rows cover the facts that decide whether it fits your team's workflow.
- Authoring
- Create and edit skills in the Registry UI or API, with categories, tags, version tracking, and enable/disable controls.
- Skill content
- SKILL.md plus supporting text or binary files, up to 50 files per skill.
- File limits
- 5 MiB per file and 10 MiB total, with normalized relative POSIX paths validated before storage.
- Access control
- Private by default. VIEW, EDIT, DELETE, and SHARE permissions scoped per user, group, and skill.
- Import source
- Any Git provider via an app or OAuth connection. One-way: repository to Registry.
- Delivery
- The AI Skills CLI syncs approved skills to Claude Code, Codex, and GitHub Copilot.
Rollout
From Repository to Developer Desktop
Three steps take a team from skills scattered across repositories to a governed catalog delivered to every developer tool.
- Step 01
Connect Git
Authorize your Git provider through an app or OAuth connection and configure the repository paths used for skill discovery.
- Resolve the selected ref to a commit
- Discover folders containing SKILL.md
- Review the per-skill import report
- Step 02
Govern Access
Keep skills private by default and grant access per user, group, and skill resource.
- Assign VIEW, EDIT, DELETE, or SHARE
- Organize with categories and tags
- Enable or disable skills as needed
- Step 03
Distribute to Developers
Use the AI Skills CLI to sync accessible, approved skills into personal or project environments.
- Claude Code
- Codex
- GitHub Copilot
Jarvis Registry
Explore Jarvis Registry One registry for skills, agents, and MCP servers
The Skill Gateway is part of Jarvis Registry, the customer-controlled control plane for governing enterprise AI skills, agents, and tools.
Ready to Govern Your Organization's AI Skills?
See how the Skill Gateway keeps approved skills private, governed, and consistently delivered to every developer tool.
Available on AWS MarketplaceFAQ
Skill Gateway questions we are asked most
These come up when platform teams evaluate how private skills are imported, governed, and delivered.
No. Import runs in one direction, from repository to Registry. The Registry reads the repository through an app or OAuth connection and never writes back.
Each discovered skill is parsed and validated before the batch is applied atomically. Item-level errors are retained, so operators can see which skills were skipped or failed without losing the overall job report.
Skills are private by default. The creator receives full permissions, and other users receive only the access explicitly granted to them. Listing and content endpoints filter by ACL.
The AI Skills CLI syncs accessible, approved skills into personal or project environments for Claude Code, Codex, and GitHub Copilot.
