AI Coding Live News

At 19:23 UTC on 2026-10-23, the official Grok Imagine account (@imagine) posted Episode 1 of its Odyssey series (post), saying it was made with Grok Imagine in a creative collaboration with Wonder Studios (@WonderStudiosX) and director Dev Walker (@glumlot_). The video runs about 7 minutes 12 seconds and is served in a 3840x2160 version. When we pulled it, the post had about 12,055 likes, 1,275 reposts, 433 quotes, 974 replies and 2.4M views, and the account asked whether people want a full breakdown of how it was made. Discussion on X kept growing over the past half day: supporters see it as AI video moving from short clips to whole episodes, while critics point to uneven acting and visible AI artifacts. This is an X discussion presented as opinions, not a verdict on the film's quality.

Claude Code v2.1.302 (Oct 16) makes fork subagents the default in claude -p, the Agent SDK, desktop and VS Code, starts permission-mode-less headless sessions on third-party providers in default mode, and adds CLAUDE_CODE_AUTO_MODE_MODEL. It also flags that standalone Claude Design (claude.ai/design) closes on Dec 14, 2026; design systems must be migrated to Claude Artifacts, while chats and comments are not carried over.

Continual learning without weight fine-tuning represents a frontier in LLM agent research. When an agent acts purely by writing and executing Python scripts inside an execution sandbox, every thought, hypothesis, and plan leaves a verifiable trace on disk. In a newly released study (arXiv:2610.11450), researchers subject coding agents to ARC-AGI-3—a suite of instruction-free abstract visual reasoning environments where each sequential level introduces novel causal dynamics. Developing a fine-grained measurement protocol, the authors trace thoughts across file artifacts across seven full evaluation runs spanning three foundation backbones from two major model families. The empirical findings reveal that scripts written for one level are almost never invoked in subsequent levels (only 33 of 630 cross-task references survive), as agents deliberately abandon 74% of scripts at task boundaries to purge level-specific overfitted state while rewriting core rules. Notes are maintained in an append-only fashion, with conflicting claims arbitrated against raw execution logs to achieve selective rather than catastrophic forgetting. Crucially, the study identifies carrying hard-coded variables across task boundaries as the single most devastating failure mode of autonomous agents.

Tackling long-horizon problem solving and scientific discovery requires computation to compound across successive attempts. While humans build upon partial proofs, failed experiments, and prior hypotheses, current test-time compute paradigms (e.g., MCTS and Best-of-N) isolate search trajectories into independent silos, discarding intermediate reasoning artifacts upon branch failure. Researchers introduce compounding inference and its open-source instantiation GitSwarm (arXiv:2610.04862). GitSwarm establishes an asynchronous multi-agent coordination architecture where homogeneous LLM agents collaborate through structured persistent memory hosted in a shared, branchable Git repository. Intermediate artifacts are preserved as atomic commits, while explicit semantic dependencies link cross-branch contributions, enabling agents to inspect, branch, verify, and synthesize past discoveries. Evaluated on rigorous long-horizon benchmarks, GitSwarm paired with GPT-5.5 solves all 30 problems on IMOProofBench-Advanced in a single execution run, and scores 79.4% on ProgramBench compared to 65.1% for the strongest reported baseline under identical budgets.

As autonomous agents execute complex multi-step workflows across terminal consoles and web browsers, context window expansion and escalating token costs impose severe operational bottlenecks. Conventional compaction methods rely on text summarization or token pruning, both of which discard crucial state variables and induce semantic drift over extended sessions. Researchers from HKUST and Alibaba present REMORY (arXiv:2610.11287), introducing a residual memory paradigm designed for sequence-level context compaction. REMORY combines lossy natural language summaries with bounded sets of learned soft memory tokens, effectively forming a residual connection in the sequence dimension. While text summaries convey high-level narrative continuity, the soft memory tokens reconstruct lost fine-grained execution artifacts, API arguments, and intermediate environment states. On challenging benchmarks including BrowseComp and Terminal-Bench 2.1, REMORY retains 97% of full-context task performance while occupying only 5.2% of the original input token footprint, slashing redundant tool retries and loops by more than 40%.

Long-horizon coding agents tackling complex software repositories depend on timely guidance, yet conventional LLM critics often degrade agent outcomes by misinterpreting in-progress exploratory commands or offering ephemeral advice that is never tracked. Researchers from Rutgers and Lehigh University unveil Opera (arXiv:2609.33987), an open-source verbal critic framework that treats diagnostic corrections as persistent notes tracked until root problems are definitively resolved. Opera orchestrates reviews through hybrid periodic and event-driven triggers, verifies proposed critiques against visible terminal artifacts, and monitors downstream trajectories to distinguish mere superficial compliance from genuine resolution. As a test-time intervention, Opera elevates baseline agent resolve rates by 12.4 percentage points on Terminal-Bench 2.1, 15.0 on SWE-Bench Pro, and 8.9 on DeepSWE v1.1 across four foundation backbones. Furthermore, fine-tuning Qwen3.5-9B on Opera-guided rollouts yields a 10.2 percentage point gain on held-out repositories without any test-time critic, preserving stability across heterogeneous execution harnesses (such as OpenHands to Terminus-2).

@DanKornas ships brief (github.com/git-pkgs/brief): an MIT single-binary CLI for contributors and AI coding agents that detects language, package manager, test/build/lint/formatter conventions; accepts local paths, Git URLs, and registry packages; can narrow reports to changed files; emits human or JSON output; and adds inspection commands for native artifacts, tooling gaps, and threat categories.

Cursor introduced Projects, replacing per-task chats with a coordinator agent in one persistent thread. The agent stays on, proactively manages work with subagents, and can set reminders, run scheduled tasks, follow PRs to fix CI, and watch Slack for bugs. Agents in a project share memory and artifacts such as plans and demos, synced across your device and agent machines. Available in beta and rolling out starting September 10.

Anthropic rolled out a major unified workspace redesign merging Claude Chat and Cowork. Users can now fluidly transition between conversational chat, persistent Cowork pairing, interactive Artifacts, and Design preview modes within a single continuous context thread.

Google AI Studio launched an updated Managed Agents harness in the Gemini API. antigravity-preview-09-2026 is live in the Interactions API and AI Studio, giving agents isolated Linux execution, filesystem tools, web search, persistent environments, and MCP connectivity. The release also adds Files and Credentials APIs so developers can move artifacts between turns and connect external services without putting secrets in the model context.

Anthropic is merging Claude Cowork and chat into one Claude experience: ask a quick question or hand off a report, keep work running after you close the laptop, and Claude asks when unclear while you keep final say. Existing Cowork chats, projects, artifacts, connectors, and skills carry over; Claude Design, Slides, and Docs are in beta on paid plans, rolling out to Pro and Max over the next few weeks.