A developer argues that "context engineering" should be understood as a per-turn packing problem rather than a synonym for prompt writing, framing the LLM context window as a fixed-token cache with re…
A developer's field notes argue that many LLM tasks are over-engineered as agents, and instead propose three fixed pipeline shapes—sequential steps, single classification, and parallel fan-out—that av…
A developer from Loop & Retry explains that streaming tool calls from LLM APIs like Anthropic's requires treating partial JSON fragments differently from streaming text, and recommends a tolerant pars…
A developer's blog post on Loop & Retry argues that most code retries by default without considering when to give up, and outlines decision rules for user-facing versus background operations. The post…
A developer's blog post on Loop & Retry details how to build observability into retry decisions for LLM agents in production. The post emphasizes that retry attempts are indistinguishable between tran…
Loop & Retry, a developer-focused publication, outlines a pre-release testing strategy for LLM agents that predicts production failures by deliberately injecting faults and testing on real input distr…
A developer's blog post on Loop & Retry highlights a hidden failure mode in LLM agent fleets: orphaned retries that continue working after the caller has abandoned the request. The post proposes treat…
A developer's field notes on multi-agent LLM systems identify coordination failures that single agents cannot exhibit, including correlated errors that amplify false consensus, the bystander effect in…
An engineer from Loop & Retry details how LLM agents fail silently in production, unlike traditional services that throw exceptions. The post introduces a taxonomy of failure outcomes—success, hard_er…
A developer's analysis of LLM agent failures reveals that per-step error rates underestimate real-world risk because errors are coupled through the agent's context, causing cascades that amplify a sin…
A developer from Loop & Retry argues that most LLM agent bugs stem from poorly designed tools rather than the model itself, and outlines four properties—legible schemas, validating boundaries, recover…
An engineer from Loop & Retry argues that debugging a failed LLM agent run is far more expensive than the run itself, because agent failures are stochastic and non-reproducible, forcing engineers to p…
A developer's analysis of best-of-N sampling for LLM agents shows that firing N parallel attempts costs 2.9x the tokens of sequential retry for the same success rate, because best-of-N pays for all at…