OpenCowork v0.14.13

OpenCowork v0.14.13 Release Notes

Item	Value
Version	v0.14.13
Date	2026-05-21
Release type	Scheduler reliability and scheduled-delivery release
Previous tag	v0.14.12

Release Focus

v0.14.13 is the current public release line for OpenCowork. It completes a recent scheduler-focused release slice that started in v0.14.10: scheduled runs should execute the intended work directly, compute the next automatic run accurately, retry when configured, and explicitly deliver text or files back to Feishu when the workflow requires it.

This is an open-source evaluation release for builders, researchers, operators, and product teams exploring local-first AI workflow automation. It does not claim hosted SaaS availability, enterprise compliance, production certification, guaranteed automation success, or managed support SLAs.

Highlights

• Added runtime guardrails so scheduler-triggered runs execute the current task instead of recursively creating or editing scheduler jobs.
• Added explicit send_feishu_message delivery and expanded attachment delivery so scheduler and other non-IM runs can send results to the currently bound Feishu account.
• Replaced placeholder cron next-run calculation with timezone-aware parsing through cron-parser.
• Updated scheduler refresh behavior so manual triggers do not overwrite the next automatic execution time incorrectly.
• Added retry-aware scheduler failure handling that requeues eligible tasks with the configured retry delay.
• Updated the queue loop so delayed tasks wake up at the correct time without unnecessary busy polling.

Why It Matters

OpenCowork is increasingly useful for recurring local workflows such as scheduled research checks, routine browser-driven reporting, IM-delivered result summaries, and repeatable template execution. Those workflows become hard to trust when scheduled runs misread their prompt, show misleading next-run times, fail once without retry, or cannot return results to the right user.

v0.14.13 improves that operational baseline without changing the project's public positioning: OpenCowork remains a local-first, contributor-friendly runtime for evaluation and prototyping, not a production-certified automation cloud.

Best-Fit Evaluation Scenarios

• Scheduled browser research or monitoring on a dedicated local desktop or VM.
• Template-driven recurring tasks that need accurate next-run visibility.
• Feishu-connected workflows where scheduled runs should send summaries or files back to a bound user.
• Contributor testing of scheduler, IM, and runtime behavior before broader productization.

Open-Source Marketing Notes

• Describe OpenCowork as an open-source, local-first desktop agent runtime for evaluation, experimentation, and workflow prototyping.
• Emphasize observable task runs, reusable templates, scheduler support, MCP interoperability, and human oversight.
• Avoid language that implies hosted multi-tenant SaaS, production certification, compliance guarantees, commercial pricing, or service-level commitments.

Recommended Usage Notes

• Run scheduled workflows on a trusted local machine, VM, or dedicated automation device.
• Keep config/llm.json, config/feishu.json, tokens, cookies, generated databases, and private task artifacts out of Git.
• For scheduled tasks that must notify a user, make sure the device is bound to Feishu and the workflow explicitly sends a message or attachment.
• Validate recurring workflows manually before leaving them unattended.

Verification

npm run build
npx vitest run src/scheduler/cronParser.test.ts src/scheduler/scheduler.test.ts src/scheduler/taskQueue.test.ts src/scheduler/taskExecutor.test.ts src/im/DispatchService.test.ts src/im/feishuPrompt.test.ts

Publishing Checklist

1. Update public docs and changelog entries for v0.14.10 to v0.14.13.
2. Use existing tag v0.14.13.
3. Push the docs update to main.
4. Create the GitHub Release for v0.14.13 using this file as the release body.

GitHub release publishing requires gh authentication or a GH_TOKEN with release permissions.