Topic · 36 stories · page 1 of 2
36 stories tagged ai-agents on AI Tamer, newest first.
AWS Continuum for code vulnerabilities is generally available in two US regions. It sandboxes exploits, ranks by business impact and opens fix PRs you approve.
Muse, Meta's personal AI agent, is live in Japan on iOS, Android and muse.ai. Mercari, Money Forward ME and Ticket Pia links are still being built.
From next week, new sessions in most Claude apps on Pro, Max and Team skip per-step approvals by default. Here is where Manual lives and who stays opt-in.
OPAL wraps a PyTorch optimizer so Claude Code can change learning rates, clipping and update rules mid-run. Single-run gains, and it can make runs worse.
Claude Duet is a waitlist preview for US Pro and Max users. Claude Security now rescans repos as code changes, in beta for Enterprise. What each does.
Microsoft says a China-aligned group used ARTEX agents after the ConnectWise Cloud Backup breach, trying 2,250+ Microsoft 365 tenants. What to check.
Claroc says one passive-DNS query listed 6,800+ Muse share hostnames, each named after its chat. Some it opened held loan, revenue and job-search details.
Yelp, Delta, JPMorgan Chase, and Charles Schwab are on record limiting consumer AI agents. Other blocks in the same reports are user anecdotes.
CERT-In says unnamed AI providers disclosed that agents reached Indian organisations' live systems and read personal and confidential records.
A 14 October report says agents on Tencent Cloud made more than 2,600 attempts to pull Amap data. Which model ran is unresolved.
An agent that retries a failing tool call immediately, at every layer, can turn a short blip into an outage. Backoff, jitter, retry limits and which errors to retry at all.
If your agent logs one JSON object per line, five short jq filters turn that log into answers: which tool calls failed, which tools fail most, which steps were slowest, and how much time the logged calls span.
A Linux TCP connection can look healthy while an AI request stalls. Keepalive, user timeout, and an application deadline answer different questions.
Python asyncio.shield protects an inner task from cancellation propagated by its waiter. The caller still receives CancelledError and must account for work that continues.
A keypad press arrives as a distinct Twilio WebSocket event in a bidirectional voice stream. Route it separately from speech before turning either input into an application command.
The Wikimedia Foundation says agents it believes OpenAI operated made sandbox edits, probed a hosted note tool and sent heavy traffic. It found no evidence of compromise or coordination on its systems.
Head sampling controls trace volume early. Tail sampling can retain runs that later fail or become slow, if the pipeline can hold and route their spans.
Aviram explains why separate credentials, limited permissions and protected backups matter when AI agents work with real infrastructure.
"The agent did it" is where an incident review starts. What to record in the timeline when an AI agent took an action, and a list of contributing factors that points at the system around it.
A pipeline, a log tee or a local variable can turn a failed command into exit status 0. Patterns that hide failure from CI and AI agents, and how to make the status tell the truth.
Cron jobs overlap, CI retries, queues redeliver and AI agents try again when unsure. A practical guide to writing scripts whose second run changes nothing and whose half-finished run can simply be repeated.
API keys rarely leak through the vault. They leak through a command line other users can read, a URL that gets logged, and a conversation transcript that gets archived.
When AI agents send each other messages, the old queueing problems come back with new twists: a message delivered but never read, a duplicate that repeats an action, and a relayed yes that nobody actually gave.
Agent logs are tempting to fill with everything: prompts, tool calls, environment, full requests. A short list of what must never land in them, and why scrubbing has to happen before the write.
