Tech News
← Home  ·  All topics

Agent

159 GoKawiil briefs on this topic

Parseable's Data Lake Processes 100 Million Time-Series Data Points per Minute

Parseable has announced its open observability data lake capable of handling up to 100 million time-series data points every minute. The platform leverages large language models and traditional machine learning to analyze telemetry data, detect anomalies, and identify root causes in real-time. Recently, it identified a spike in error rates caused by exhausted connection pools in a checkout service, affecting over 6,400 users for more than seven minutes.

Innovative ephemeral testing method uses AI to evaluate software layers

A developer introduces ephemeral testing, a new approach where AI builds and tests temporary software layers on top of existing code, which are discarded after evaluation. This process emphasizes assessing the quality of the integrated system rather than the original code directly. The method relies on stable, well-documented libraries to produce meaningful feedback through failures and patches.

Meta patches critical VM escape flaws in Muse AI before launch

Ahead of Muse's release, Meta engineers identified multiple security vulnerabilities in the AI agent's virtual machine environment. At least one flaw could have enabled malicious users to access Meta's sensitive internal data, prompting urgent fixes involving multiple teams and senior management. The vulnerabilities were linked to Linux kernel virtualization code and required immediate attention to prevent potential data breaches.

Companies shift focus from AI task success to overall business improvement

Many organizations are moving beyond basic automation to integrate AI in a way that enhances overall company performance. Instead of just measuring whether individual AI agents succeed at specific tasks, they are now seeking ways to evaluate how AI contributes to broader strategic goals. This transition reflects a desire to leverage AI as a tool for meaningful business growth.

Apple to restrict Full Disk Access on Mac after Meta Muse privacy concerns

Apple said it will add new controls around macOS's Full Disk Access permission following reports that Meta's Muse AI agent used that access to build detailed profiles of users' contacts, tracking relationships, locations and personal milestones on an hourly basis. Security researcher Karan Joshi reportedly found Muse was instructed to compile data such as where people live, recurring life events and significant dates for everyone in a user's life.

Meta's Muse AI assistant found to build detailed profiles of users' contacts

Researchers including Karan Joshi extracted Muse's internal system prompts by asking the chatbot to reveal its own files, exposing instructions that direct it to create a 'page for every person' in a user's life. The process reportedly runs hourly, compiling data on family, partners, friends, colleagues and people a user follows into memory files Muse uses to make suggestions.

Developer's AI agent clears WoW orc starter zone using GPT-6 Astra model, no deaths

An independent developer built agent-wow, an open-source client for AzerothCore private servers, and used it with OpenAI's GPT-6 Astra model to complete every quest in World of Warcraft's Valley of Trials zone in 40 minutes without dying. The AI never saw rendered game visuals, instead reading raw network protocol messages and server quest data to decide its actions via a Python script. The developer says the long-term goal is populating a private server with many such agents to attempt clearing Icecrown Citadel on heroic difficulty.

GitHub makes redesigned dashboard the default for all users

GitHub has rolled out its redesigned dashboard to all users by default, ending its status as an opt-in feature preview. The new layout consolidates active agent sessions, issues and pull requests into filterable lists of up to 12 items each, separates general updates into a dedicated Feed tab, and lets users start tasks directly from the dashboard, such as assigning issues to Copilot coding agent. Users who prefer the old layout can revert via a dropdown next to 'Preview.'

Offrun launches as unified dashboard for managing multiple coding agents

A developer has released Offrun, a tool that consolidates management of multiple AI coding agents into a single workspace. The Show HN post includes an example debugging scenario: an upload test that intermittently failed because it checked a file before the upload had finished, fixed by waiting on the upload process itself rather than a fixed timer.

New open-source tool 'pi pod' adds sandboxing to the pi coding agent

A developer has released pi pod, a self-hostable project that runs the pi coding agent inside isolated sandboxes on a user's own server. It adds features like sandboxing, native clients, session sharing with role-based access, and browser/app automation on top of the existing pi harness. The creator says a hosted version of the service is planned for later release.

Developer's agent-wow lets GPT-6 Astra complete WoW starting zone quests

A developer built an open-source project called agent-wow to test frontier language models inside World of Warcraft, using reasoning rather than game-specific training. Running Codex with GPT-6 Astra (xhigh), the agent created an orc character and completed all starting-zone quests in 40 minutes with zero deaths and minimal issues.

GitLab patches critical AI Gateway flaw allowing remote code execution

GitLab issued an advisory urging immediate patching of CVE-2026-90970, a critical vulnerability in its AI Gateway service that powers GitLab Duo. The flaw lets an authenticated user with Duo Agent Platform access escape a prompt template sandbox via a crafted flow configuration and run arbitrary commands on self-hosted AI Gateway instances. GitLab released versions 19.2.4, 19.3.2, and 19.4.1 to fix the issue, noting GitLab-hosted AI Gateway customers are already protected.