| name | flyio |
| id | flyio |
| description | Fly.io integration for application monitoring, machine lifecycle management, metrics, logs, and incident remediation |
| category | cloud_provider |
| connection_check | {"method":"is_connected_function","module":"chat.backend.agent.tools.flyio_tool","function":"is_flyio_connected"} |
| tools | ["cloud_exec","query_flyio_metrics"] |
| index | Fly.io — apps, machines, logs, metrics, deployments, health checks, remediation |
| rca_priority | 8 |
| allowed-tools | cloud_exec, query_flyio_metrics |
| metadata | {"author":"aurora","version":"1.0"} |
Fly.io Integration
Overview
Fly.io is connected for application monitoring, machine lifecycle management, and incident remediation.
Instructions
How to interact
- Use
cloud_exec('flyio', '<fly command>') for all Fly.io CLI operations.
- The CLI (
flyctl) is pre-authenticated with the user's org-scoped token.
- Always pass
--json flag for structured output when available.
Prometheus metrics
Use query_flyio_metrics(query) for PromQL queries against Fly.io's Prometheus federation. Common metrics include (but are not limited to) fly_instance_up, fly_instance_cpu, fly_instance_memory_resident, fly_edge_http_responses_count, fly_edge_http_response_time_seconds_bucket, fly_instance_net_recv_bytes, fly_app_concurrency.
Critical rules
- Always use
cloud_exec('flyio', ...) -- never call the REST API directly.
- Use
-a <app_name> for commands that target a specific app (e.g. fly status -a myapp). Do NOT use -a on global commands like fly apps list, fly regions list.
- Pass
--json for structured output when available (not all commands support it -- if it fails, retry without).
- For logs, always use
--no-tail to avoid indefinite streaming (e.g. fly logs -a myapp --no-tail).
