spec_version: 1
name: disk_panic
description: "SRE incident response RL environment — fix a production server with a full disk and crashing services without touching business-critical logs."
version: 0.1.0
author: yashppawar
type: space
runtime: fastapi
app: disk_panic.server.app:app
port: 8000
tasks:
  - id: easy
    description: "Free the disk by removing a bloated rotated log."
  - id: medium
    description: "Free the disk, restart the crashed app.service, preserve /var/log/audit."
  - id: hard
    description: "All of the above + tame a runaway writer via a logrotate config."
action_space:
  type: DiskPanicAction
  fields:
    command: "str — a bash-like command (ls, df, du, cat, rm, find, sha256sum, systemctl, or `echo ... > path`)"
observation_space:
  type: DiskPanicObservation
  fields:
    stdout: "str — output of the last command"
    df_output: "str — simulated `df -h /` output"
    service_status: "str — simulated systemctl state of app.service"
    task_id: "str — current task (easy/medium/hard)"
    step: "int"