add contribution part

Author: Leolty

blog.html

@@ -266,9 +266,15 @@ <h2>The CocoaAgent Framework</h2>
             To enable rigorous evaluation and empower researchers to develop their own agents, we built the <strong><a href="https://github.com/cocoabench/cocoa-agent">CocoaAgent</a></strong> framework. CocoaAgent provides seamless integration with <a href="https://github.com/agent-infra/sandbox">AIO Sandbox</a>, an all-in-one Docker environment. It equips agents with a full suite of tools—browser automation, terminal access, file operations, and code interpreters—enabling them to operate like human developers in realistic settings. Our framework is model-agnostic, and we provide example scripts for running agents with both open-source LLMs such as Qwen3-VL and commercial models such as GPT-5.1. To support robust evaluation at scale, CocoaAgent implements both dynamic runtime tests for verifying computational correctness and lightweight static-matching checks for deterministic answers. We are currently producing comprehensive results on CocoaBench using this framework, which will be available soon on our leaderboard.
         </p>
 
-        <h2>Future Work</h2>
+        <h2>Get Involved</h2>
         <p>
-            We are continuously building and improving CocoaBench. Collaboration is welcomed! CocoaBench benefits from collaborators with diverse backgrounds. We invite you to bring your interesting ideas and use cases for agents to the community. Feel free to reach out to us or join our <a href="https://discord.gg/ZDaDhVCd" target="_blank">Discord community</a> to propose new tasks or discuss ideas. We will be also setting up a task contribution protocol soon.
+            We are continuously building and improving CocoaBench. CocoaBench is a community-driven benchmark, and we welcome contributions from researchers and practitioners with diverse backgrounds. If you've encountered a challenging real-world problem that pushed your limits, it might make a great benchmark task!
+        </p>
+        <p>
+            We've set up a streamlined <a href="https://github.com/cocoabench/cocoa-agent/blob/main/contrib/CONTRIBUTING.md" target="_blank">task contribution protocol</a> to guide you through creating and submitting new tasks. Contributors with <strong>3 accepted tasks</strong> are eligible for co-authorship on the CocoaBench paper, which we plan to submit to a top-tier ML conference.
+        </p>
+        <p>
+            Have questions or ideas? Feel free to reach out to us or join our <a href="https://discord.gg/ZDaDhVCd" target="_blank">Discord community</a> to propose new tasks or discuss ideas. (If the link doesn't work, try refreshing the page or manually add the server in Discord app using invite code: <code>ZDaDhVCd</code>)
         </p>
 
         <hr style="border: 0; border-top: 1px solid #eee; margin: 40px 0;">