traces
7 stories and discussions about traces, aggregated from every source we track.
A Kubernetes tool leveraging eBPF for advanced Kubernetes security, auto-generating Network Policies, Seccomp Profiles, and more. - kguardian-dev/kguardian
Asynchronous monitoring, incident investigations, and compliance audits primarily rely on agent traces to reconstruct what happened. These analyses assume that LLM agents cannot tamper with their own execution traces.…
The most repeated number in enterprise AI turns out to be a paraphrase of unnamed opinion surveys, quoted in a magazine profile of a vendor.
Semantic loop detection for agent traces: a judged drop-in for deepeval's word-overlap stagnation check - VBS2004/semloop
Bar charts and findings from LLM Agents Can Easily Tamper With Their Own Traces (arXiv:2609.30266v1), covering ten model–harness pairs.
Bar charts and findings from LLM Agents Can Easily Tamper With Their Own Traces (arXiv:2609.30266v1), covering ten model–harness pairs.
Tool-using language-model agents repeatedly rediscover procedures they have already executed, producing traces that mix reusable structure with retries, exploration, accidental ordering, and repeated lookups. We…