How to use OpenClaw to run agency content workflows
A solo agency developer runs four scheduled OpenClaw agents with Gemini and local intelligence files, reporting automated content and operations alongside billing and integration failures.
news.ycombinator.com
Suggested prompt
Suggested prompt written by Agents In Use, not the source author. Replace bracketed inputs and review tool permissions first.
Using [local intelligence files], create [content deliverable] for [agency audience] in one focused request. Return [structured response format] for [publishing script]. Include a self-review against [quality criteria] and flag missing information.
Reach people exploring AI agents
See placements →Evidence at a glance
Sourced account- Source poster / publisher
- ppcvote
- Original account
- Read the source ↗
- Published
- Source link checked
- Reported result
- The author reported eight social posts daily, 27 automated Threads accounts, over 12,000 followers and over 3.3 million views, alongside billing, request-budget and Telegram failures.
- Human involvement
- The author built the system and documented operational failures. The post does not describe routine human review before each publication.
A linked account supports attribution. It does not mean we independently reproduced its result.
Report an issue with this case
Open correction emailOpens your email app with the case link included. Nothing is sent automatically.
How the agent did it
- Deployed four OpenClaw agents on WSL2 with systemd timers.
- Gathered research through RSS, Hacker News and web scraping.
- Supplied local intelligence files to focused Gemini requests.
A solo developer in Taiwan described four OpenClaw agents handling content, sales leads, security scanning and operations for a technology agency. The agents run on WSL2 at home, using Gemini 2.5 Flash and 25 systemd timers. The author reported roughly 105 daily requests against a stated free-tier allowance of 1,500.
Their request pattern avoids long conversations. Each request uses precomputed local intelligence files, one focused prompt with context injected, and a response that scripts parse before acting. Research collection through RSS, Hacker News and web scraping uses HTTP and Jina Reader rather than LLM calls. Creative and analytical work goes to the model.
The author reported eight generated social posts daily and a pipeline that also covers bilingual site content, blog publishing and Discord notifications. They cited 27 automated Threads accounts, more than 12,000 followers and over 3.3 million views.
The account includes failures: a billing-enabled API key produced a $127 bill, an engagement loop consumed 800 requests in one day, and a Telegram health check conflicted with gateway polling.
Who it's for: Solo operators comfortable maintaining scheduled scripts and monitoring usage across several automated business workflows.
Prompt
Suggested prompt (written by Agents In Use, not the author's)
Source: Hacker News · ppcvote
Frequently asked questions
How does the workflow limit model conversations?
The author uses precomputed intelligence files, a single focused prompt and one response that scripts parse and act on. Research gathering uses HTTP and Jina Reader without LLM calls.
Was the setup always free of model charges?
No. Although the author reported current monthly LLM cost of $0, they also described a $127 bill in seven days from a key created in a billing-enabled GCP project.
More details
Remaining steps
- Parsed model responses and acted through scheduled scripts.
- Published content and issued notifications through the pipeline.
About this account
Scheduled agents handle recurring creative and analytical work, with scripts gathering research, parsing responses and carrying out subsequent actions.
Platform: Hacker News. Attribution: https://news.ycombinator.com/item?id=47296664. Prompt: Suggested (written by Agents In Use, not the author's). Sourced by the Agents In Use pipeline.
Credit: ppcvote on Hacker News
Source published
Added to Agents In Use
Source checked
Tools and topics
OpenClaw · Gemini 2.5 Flash · WSL2 · systemd · Jina Reader · Telegram Bot · Discord
Content automation · Agency operations · Scheduled agents
Editorial completeness: 100/100. Measures documented fields, not agent performance or verified outcomes.