Integrations
Arness connects to external services through MCP servers and CLI tools. All integrations are optional — Arness detects what’s available during init and adapts accordingly.
Platform CLIs
Section titled “Platform CLIs”| Service | CLI | Used by | Setup |
|---|---|---|---|
| GitHub | gh | Code, Infra — PRs, issues, CI status | gh auth login |
| Bitbucket | bkt | Code, Infra — PRs, issues, pipelines | bkt auth login |
GitHub and Bitbucket integration uses their native CLIs rather than MCP — no additional setup beyond authenticating the CLI.
MCP Servers
Section titled “MCP Servers”| Service | MCP Server | Used by | What it enables |
|---|---|---|---|
| Jira Cloud | Atlassian MCP | Code | Issue creation, backlog browsing, PR review tracking |
| Figma | Figma MCP | Spark | Style exploration, design grounding for prototypes |
| Canva | Canva MCP | Spark | Asset export, brand grounding for prototypes |
MCP servers are configured at the project level in .mcp.json, not baked into the plugin.
Cloud Providers (Infra)
Section titled “Cloud Providers (Infra)”Arness Infra supports infrastructure generation for:
| Provider | IaC Tools |
|---|---|
| AWS | OpenTofu/Terraform, Pulumi, CDK |
| GCP | OpenTofu/Terraform, Pulumi |
| Azure | OpenTofu/Terraform, Pulumi, Bicep |
| Fly.io | fly CLI, OpenTofu/Terraform |
| Railway | railway CLI |
| Vercel | vercel CLI |
| Netlify | netlify CLI |
| DigitalOcean | OpenTofu/Terraform |
| Render | render CLI |
| Cloudflare | wrangler CLI |
| Heroku | heroku CLI |
| Scaleway | OpenTofu/Terraform |
Infra also supports Kubernetes/Helm for container orchestration across any provider.
How Detection Works
Section titled “How Detection Works”During initialization (/arn-code-init, /arn-spark-init, or /arn-infra-init), Arness:
- Checks for Git — detects repository status and remote configuration
- Detects platform — identifies GitHub or Bitbucket from the remote URL
- Checks CLI auth — verifies
gh auth statusorbkt auth status - Scans for MCP servers — reads
.mcp.jsonfor configured services - Records findings — stores detected capabilities in the
## Arnessconfig section
Skills adapt their behavior based on detected capabilities. For example:
- If no issue tracker is detected, issue-related features are skipped gracefully
- If Figma MCP is available, Spark’s style exploration can ground designs from Figma files
- If no cloud CLI is installed, Infra suggests installing one during the wizard
Adding an Integration Later
Section titled “Adding an Integration Later”If you install a CLI or configure an MCP server after init, just run init again:
/arn-code-init # Re-detects platform capabilities/arn-spark-init # Re-detects design tool integrations/arn-infra-init # Re-detects cloud CLIs and toolsInit is non-destructive — it preserves existing configuration and adds newly detected capabilities.