Browse, install, and manage integration skills that give your agents new capabilities.
The Community Skills Marketplace is where you discover and install integration packages that give your agents new capabilities. Each skill defines a set of tools — like "create GitHub issue," "send Slack message," or "query Salesforce contacts" — that agents can use once the skill is installed and connected.
The marketplace has four tabs:
| Tab | Shows |
|---|---|
| Browse | All available skills, searchable and filterable |
| Installed | Skills you've installed — manage credentials, enable/disable, uninstall |
| Featured | Editor's picks and popular skills |
| Updates | Available updates for installed skills, update history |
Each skill card shows:
Click any skill card to view full details including tool list, tags, reviews, and repository link.
Click Connect on any skill card to open the credential setup modal. Depending on the skill's authentication type, you'll see different options:
Two options (tabbed interface):
For OAuth App setup, use this redirect URI in the service's developer portal:
https://your-domain.com/api/engine/oauth/callback
Paste the API key or token from the service's settings page. One field, encrypted with AES-256-GCM.
Some services require multiple values (e.g., AWS Access Key ID + Secret Key). Each field is labeled and encrypted separately.
When connecting a skill, you choose the credential scope:
| Scope | Behavior | Use When |
|---|---|---|
| Organization-wide | One set of credentials shared by all agents | Most cases — shared Slack bot, company GitHub token |
| Per-Agent | Each agent gets its own credentials | Different permission levels per agent (read-only vs. write) |
The Updates tab manages the lifecycle of installed skills.
| Setting | Options | Default |
|---|---|---|
| Auto-Update | Enabled / Disabled | Disabled |
| Check Interval | Hourly, Daily, Weekly, Manual Only | Daily |
| Max Risk Level | Low, Medium, High, Critical | Medium |
When Max Risk Level is set to "Medium," auto-updates will only apply updates that don't increase risk above medium. Higher-risk updates require manual approval.
Each update shows the version change (e.g., v1.2 → v1.3), any risk level changes, and a changelog when available.
The update history table shows all past updates with skill name, version change, status (applied/failed/skipped), date, and who applied it.
You can import custom skills from GitHub repositories:
agenticmail-skill.json manifestTo publish a skill to the marketplace, create a repository with an agenticmail-skill.json manifest that defines:
Each skill has a risk assessment that indicates its potential impact:
| Level | Meaning | Examples |
|---|---|---|
| Low | Read-only access, no sensitive data | Weather API, public data queries |
| Medium | Write access to non-critical systems | Slack messaging, calendar events |
| High | Write access to critical systems | GitHub code changes, database writes |
| Critical | Can cause significant damage if misused | Payment processing, infrastructure changes |
| Issue | Solution |
|---|---|
| Skill installed but tools not available | Credentials not configured. Click Connect on the skill in the Installed tab. |
| "Connected" but agent can't use tools | The skill may be disabled for that agent. Check the agent's Skills tab. |
| OAuth popup closes without connecting | Check browser popup blocker. Ensure redirect URI matches. Try the Access Token approach instead. |
| GitHub import fails | Verify the repository has an agenticmail-skill.json file at the root. The repo must be public or accessible. |
| Auto-update not working | Check Update Configuration — auto-update may be disabled or interval set to "Manual Only." |
| Skill shows "Update Available" dot | Go to the Updates tab to review and apply the update, or skip it. |
AgenticMail Enterprise — Community Skills Marketplace Documentation