Local Agent Setup: ClickUp + Qwen 2.5 (14b) | Context Stability & Gateway Timeout Issues
Sure! Since the community you showed is primarily English-speaking, here is the perfect post based on the Posting Guidelines and your specific technical setup:
Title:
Local Agent Setup: ClickUp + Qwen 2.5 (14b) | Context Stability & Gateway Timeout Issues
Message:
Hi everyone!
I’m currently deploying a Self-Hosted Autonomous Agent using the OpenClaw 2026 framework. My goal is to delegate ClickUp tasks to the agent so it can autonomously execute app prototypes and n8n workflows.
My Setup:
Hosting: Ubuntu 24.04 LTS (Local machine).
Model: qwen2.5-coder:14b running via Ollama.
Integrations: ClickUp (specific Task Lists), Supabase (for step-by-step logging), and Discord for status announcements.
What I'm trying to do:
I have an OpenClaw Cron Job set to run every hour. The agent is supposed to read pending tasks from ClickUp, decide on the best tools (Bash, Python, or n8n), execute the technical action, and log the result to Supabase.
The Challenges (Where it gets "wild"):
Gateway Timeouts: I frequently hit Error: gateway timeout after 30000ms. Since I'm running a 14b model locally, the inference time seems to exceed the default gateway wait time.
Context Instability: After memory compaction, the agent often fails to re-read essential startup files like WORKFLOW_AUTO.md or the current daily memory file. It ends up in a loop trying to recreate its own operating protocols.
"Ghost Work": I've noticed the agent sometimes logs that a task is "completed" or "in progress" without actually executing the underlying bash/python commands.
Questions for the crew:
Has anyone found a way to reliably increase the global gateway timeout for heavier local models?
Any tips on forcing the agent to prioritize "Reading" its WORKFLOW_AUTO.md immediately after a context reset?
Is anyone else seeing "lazy" behavior with Qwen where it reports success in the logs but skips the execution step?
Would love to hear from anyone dealing with similar local infra headaches!
Title:
Local Agent Setup: ClickUp + Qwen 2.5 (14b) | Context Stability & Gateway Timeout IssuesMessage:
Hi everyone!
I’m currently deploying a Self-Hosted Autonomous Agent using the OpenClaw 2026 framework. My goal is to delegate ClickUp tasks to the agent so it can autonomously execute app prototypes and n8n workflows.
My Setup:
Hosting: Ubuntu 24.04 LTS (Local machine).
Model: qwen2.5-coder:14b running via Ollama.
Integrations: ClickUp (specific Task Lists), Supabase (for step-by-step logging), and Discord for status announcements.
What I'm trying to do:
I have an OpenClaw Cron Job set to run every hour. The agent is supposed to read pending tasks from ClickUp, decide on the best tools (Bash, Python, or n8n), execute the technical action, and log the result to Supabase.
The Challenges (Where it gets "wild"):
Gateway Timeouts: I frequently hit Error: gateway timeout after 30000ms. Since I'm running a 14b model locally, the inference time seems to exceed the default gateway wait time.
Context Instability: After memory compaction, the agent often fails to re-read essential startup files like WORKFLOW_AUTO.md or the current daily memory file. It ends up in a loop trying to recreate its own operating protocols.
"Ghost Work": I've noticed the agent sometimes logs that a task is "completed" or "in progress" without actually executing the underlying bash/python commands.
Questions for the crew:
Has anyone found a way to reliably increase the global gateway timeout for heavier local models?
Any tips on forcing the agent to prioritize "Reading" its WORKFLOW_AUTO.md immediately after a context reset?
Is anyone else seeing "lazy" behavior with Qwen where it reports success in the logs but skips the execution step?
Would love to hear from anyone dealing with similar local infra headaches!
The AI that actually does things. Emails, calendar, home automation - all from your favorite chat app. New shell, ready to help. The lobster way.
134,827Members
Resources
Similar Threads
Was this page helpful?
