Find articles, news, tools and templates by subject or by the work you need to do.
91 results for “Evaluate an agent” · 25–36
News · AI in production ·
Claude’s on-demand compaction lets an application replace older conversation history with a signed summary. A conditional restart shows why the summary must preserve relationships between instructions.
Read news: Claude adds on-demand compaction for long-running agentsGuide · AI in production ·
Herdr combines agent status, terminal panes and session recovery. Its server architecture explains what continues after a disconnect and what a restart restores.
Read guide: Herdr brings coding-agent sessions into one terminalExplainer · AI in production ·
A rollback proposal crosses from recorded evidence into a changing system. The handoff carries source references, current preconditions and unfinished actions.
Read explainer: How an AIOps finding becomes an agent actionTemplate / reference · AI in production
A small deterministic example that returns pass, fail, or unknown from evaluator-owned observations.
View resource: Incident-agent recovery graderNews · AI in production ·
Anthropic cut live internet access from internal evaluations after agents acted on real systems. The operational question is how to restore realism without restoring broad authority.
Read news: Anthropic takes its internal AI evaluations offlineExplainer · AI in production ·
PulsarLM removes low-attention pages rather than summarizing them. A synthetic archive-retention case explains what a local agent’s recall trial must inspect.
Read explainer: PulsarLM keeps context small by dropping old token pagesGuide · AI in production ·
Run an AI agent with useful access to files and services, then test exactly where OpenShell stops it. A hands-on guide to sandboxes, policy, credentials, and SRE workflows.
Read guide: What is OpenShell and how do I use it?Explainer · AI in production ·
Test selective recall, corrected state, tool-failure recovery, clarification, and the independent destination result before trusting a long-running agent.
Read explainer: Muse Wonder's Active Memory needs an operational testExplainer · AI in production ·
Bind the schedule to a model, tools, input state, destination and acceptance evidence, then reconcile the actual run before retrying or declaring success.
Read explainer: Scheduled AI work needs an execution contractExplainer · AI in production ·
Kimi K3 can keep a larger incident record in view. Its practical value depends on evidence handling, agent integration and the cost of useful investigation.
Read explainer: What Kimi K3’s million-token context offers SRE teamsGuide · AI in production ·
A release-note digest and a changed maintenance event show two uses of Muse’s persistent agent: gathering information over time and preparing a connected-account update.
Read guide: Meta Muse keeps research and connected tasks runningExplainer · AI in production ·
Scion reduces repeated repository work for online agents. Its cache hierarchy explains why warm checkout speed and fleet recovery need different tests.
Read explainer: Anthropic Scion puts source control behind a cache hierarchy