OpenClaw delivers digests after a configuration fix
Rezvov reports five empty digest runs before a configuration change, followed by six articles delivered to Telegram in a thirty-second test.
blog.rezvov.com
Suggested prompt
Suggested prompt written by Agents In Use, not the source author. Replace bracketed inputs and review tool permissions first.
Investigate this digest job without assuming its delivered flag proves success. Check the cron summary and announcement input, explain any empty output, make a focused configuration correction and confirm the next digest appears in Telegram.
Reach people exploring AI agents
See placements →Evidence at a glance
Sourced account- Source author
- Alex Rezvov
- Original account
- Read the source ↗
- Published
- 25 Feb 2026
- Source link checked
- 15 Oct 2026
- Reported result
- Rezvov reports five empty digest runs before a configuration change, followed by six articles delivered to Telegram in a thirty-second test.
- Human involvement
- The operator diagnoses logs, changes configuration and checks the actual Telegram chat. The reported repair applies to his tested setup.
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
- Rezvov inspects digest jobs with empty summaries despite a delivered status.
- Rezvov traces the empty announcement input to the cron response and disables thinking in his tested configuration.
- OpenClaw reruns the RSS gathering and filtering workflow and produces six article summaries.
Alex Rezvov investigated OpenClaw article digests that appeared successful in logs but never arrived in Telegram. The cron agent gathered and filtered RSS content, yet returned an empty or introductory text response. A separate announcement agent therefore had no useful digest to forward, even though the system recorded its own clean exit as delivered.
Rezvov traced the empty output to the configured thinking mode in his setup. He reports five runs averaging 88 seconds without deliverable content. Turning thinking off produced a thirty-second run that delivered six articles on the first attempt. The account also distinguishes that fix from session isolation, which addressed a different loop involving an invented workflow filename.
The useful result is an observed delivery correction with a clear inspection point. A completed process and a delivered flag did not prove a message had reached the intended chat. Rezvov advises checking the summary field and the Telegram conversation after configuration changes. This entry preserves his particular diagnosis and reported test results, without turning it into a universal explanation for every missing digest. Model choice, context and announcement behavior can also affect the result.
Prompt
Suggested prompt (written by Agents In Use, not the author's)
Source: Blog · Alex Rezvov
Frequently asked questions
Why was delivered true misleading?
It recorded that the announcement agent exited without error, not that a Telegram message had arrived.
What changed in the successful test?
Rezvov disabled thinking for the cron agent in his setup and obtained a six-article digest.
How should delivery be checked?
Inspect the summary content and confirm that the message appears in the intended Telegram chat.
More details
Remaining steps
- Rezvov confirms the digest actually reaches Telegram instead of relying on the delivered flag.
About this account
The cron agent gathers and filters RSS articles, produces the digest and passes it to the announcement agent for Telegram delivery after the owner fixes configuration.
Firsthand account read. Primary BlogPosting datePublished 2026-02-25 matches the visible date. Configuration diagnosis and timings belong to this deployment, not every missing-message incident. Reported outcomes were not independently reproduced. Suggested prompt written by Agents In Use.
Credit: Alex Rezvov
Source published 25 Feb 2026
Source checked 15 Oct 2026
Tools and topics
OpenClaw · RSS feeds · Telegram · Agent logs
OpenClaw · RSS feeds · Telegram
Editorial completeness: 100/100. Measures documented fields, not agent performance or verified outcomes.