v0.2.0

0.2.0

The 0.2 task and integration design, promoted from 0.2.0-alpha with updated documentation and agent skills.

Highlights

• Opt-in Git snapshots and S3-compatible artifact uploads, composed with ordinary commands.
• Fetch stored artifacts with nextask s3 fetch TASK_ID --to DIR, with include/exclude filters, dry runs, and explicit overwrites.
  Fetching requires no database connection.
• Caller-supplied task IDs, worker task environment, and shared configuration with environment-based credentials.
• Consistent list flags, duration units, output, and tag filtering; corrected wait --any and stale-worker filtering.
• Confirmed daemon startup, cancellation recovery, and durable replay of journaled task results after worker loss.
• Updated setup and task-management skills with examples, verification, and troubleshooting guidance.

Upgrading

From 0.1.x, replace literal credentials with environment references or supply complete connection URLs through the environment.
S3 endpoint URLs carry access-key and secret-key references.
Run nextask init db and upgrade workers before enqueueing integrated tasks.
Config files are validated before overrides, so replace literal credentials in both global and project configs.

From 0.2.0-alpha, there are no additional schema or configuration changes.
Upgrade the CLI and workers together.

Current limits

• Worker-failure classification, automatic task requeue, and resume/restart attempts remain planned.
• A stale heartbeat does not guarantee the old command has stopped.
• Completion recovery requires a saved journal on the worker's disk; it does not resume interrupted commands.
• Abrupt instance loss preserves only completed artifact uploads; uploads are not atomic filesystem snapshots.

See upgrade instructions, integrations, and configuration for migration details and recovery limits.