AI Weekly Malaysia

Anthropic engineer Sachin Malhotra describes how Claude Tag acts as the first responder for CI/CD failures at Anthropic, publishing initial situation reports within 15 minutes of incidents. The agent holds memory across a dedicated on-call Slack channel, uses a service account with tool access to investigate, and accepts natural-language scheduling prompts like 'run CI handoff every Monday at 9:00am EST.' In one example, Claude identified that 44 missing tests were caused by a feature flag turned on that morning and recommended reverting it.

Why: If you maintain CI/CD pipelines and want to reduce on-call pain, this article outlines a concrete agent architecture—Slack channel as memory surface, service account with scoped tool access, natural-language scheduling—you can replicate with Claude or adapt to other LLM agents. The 15-minute first-analysis benchmark is a useful target to measure your own automation against.

A Register 'On Call' anecdote features a developer sent to a power station to document and fix undocumented monitoring code running on a dusty CP/M machine connected to a turbine overspeed detector. Faced with no documentation and the risk of causing a turbine overspeed or city-wide blackout, he decided the safest action was to not touch the code at all.

Why: The story is an entertaining reminder that when you inherit undocumented critical-infrastructure code with no way to verify behavior, the correct engineering decision may be to walk away rather than make changes you cannot safely test. There is no actionable technical detail, product release, or policy change here.