MCP App StoreLe meilleur de l'écosystème IA et MCP, sélectionné chaque jour.
Sources
Cursor launched Automations — a system for defining trigger-based agents that run automatically without a developer initiating them. Developers can set up agents to respond to events like PRs, test failures, or deploys with custom instructions. Moves Cursor firmly into the background agent space alongside GitHub Actions and CI-native AI tools.
OpenAI launched ChatGPT for Excel, integrating GPT-5.4-powered assistance and new financial data connectors to speed modeling and analysis for regulated finance workflows. For developers and analysts, this provides programmatic natural-language formulas, data-cleaning helpers, and secure connectors that reduce manual spreadsheet work and lower integration overhead.
OpenAI authors extend single-minus amplitude techniques to gravitons and report results derived with assistance from GPT-5.2 Pro, advancing theoretical tools for quantum gravity calculations. This is notable for researchers using symbolic and model-aided derivations — it shows LLMs aiding nontrivial analytic work and verifies a new technical result in amplitude theory.
Cursor agents are now available in IntelliJ IDEA, PyCharm, WebStorm, and other JetBrains IDEs via the Agent Client Protocol (ACP). Developers who prefer JetBrains tools can now access Cursor's agent capabilities without switching editors. Significant for the large JetBrains user base in enterprise Java and Python shops.
OpenAI announces GPT-5.3 Instant, a tuned variant focused on lower-latency, more natural conversational turns and improved everyday usefulness. This release prioritizes response smoothness and latency optimizations for real-time applications, making it attractive for developers building chat and assistant experiences.
The system card for GPT-5.3 Instant details the model’s intended uses, safety mitigations, and benchmarked capabilities, giving developers guidance on guardrails and recommended system prompts. It’s a useful technical reference for teams integrating GPT-5.3 Instant into production systems and for evaluating suitability for specific tasks.
Anthropic shipped improvements to the skill-creator tool that let skill authors verify their skills work, catch regressions, and improve trigger descriptions — without writing code. The update adds a testing layer to the Agent Skills ecosystem, making it easier to ship reliable, well-scoped skills. Directly relevant to anyone building or maintaining SKILL.md packages.
OpenAI and Amazon introduce a stateful runtime for agents in Amazon Bedrock that brings persistent orchestration, memory, and secure execution to multi-step AI workflows. This runtime aims to make agent-based applications more reliable and easier to build by preserving context and managing execution state. Developers building multi-step agents or long-running workflows on Bedrock should evaluate this for simplified orchestration and improved memory handling.
OpenAI demonstrates a bidirectional Codex + Figma workflow: generate a running UI with Codex, bring it into Figma for design refinement, then sync changes back to Codex for implementation. This closes the design-to-code loop in a way that keeps both designers and developers in their preferred tools. An interesting workflow for teams bridging design and AI-assisted development.
Cursor CEO Michael Truell outlines a third era of AI software development — moving beyond copilot-style assistance toward autonomous cloud agents that handle large tasks over long timescales. The post argues that the shift from interactive to autonomous agents is as significant as the shift from no AI to AI assistance. Foundational reading for understanding where AI coding tools are headed.
OpenAI's February threat report analyzes how malicious actors chain models with web platforms to scale harmful campaigns and evaluates detection and mitigation strategies. For developers and security engineers, the report highlights attack patterns to watch for and recommends instrumentation and model-safety practices to reduce exploitation risk. Implementing the recommended detection signals and provenance checks will help harden systems that integrate models.
Google launched an Opal agent step that lets developers compose dynamic, agentic workflows inside Opal. This provides a first-class building block for chaining model actions and orchestration, lowering friction for developers building multi-step agentic apps.
Anthropic describes integrations between Cowork and finance plugins that streamline secure data access workflows for financial applications. For developers, the post highlights plugin patterns and permission models that make it easier to connect private datasets to agent workflows while retaining controls. This is relevant to teams building regulated-agent integrations where secure plugin architectures are required.
Anthropic launched Cowork, a collaborative workspace feature that lets teams work alongside Claude together in real time, plus a plugin system for connecting Claude to internal enterprise tools. This moves Claude beyond individual productivity into shared team workflows and custom integrations. A significant product expansion for enterprise buyers.
OpenAI explains that SWE-bench Verified has become contaminated and mismeasures frontier coding progress, recommending SWE-bench Pro instead. This is significant for researchers and toolmakers building or evaluating coding models: benchmarks can mislead product decisions if they leak or poorly reflect real-world tasks. Developers should consider benchmark provenance and prefer evaluation suites with robust guardrails like SWE-bench Pro.
OpenAI details how to use Codex for long-horizon tasks — multi-step jobs that run autonomously without constant user input. The post covers structuring tasks for reliable execution, using hosted shell and background mode, and managing state across extended agent runs. Essential patterns for anyone building production Codex workflows.
Anthropic outlines how AI-assisted migration tools reduce the cost and risk of modernising legacy COBOL systems, using code understanding and automated transformation to accelerate porting. This matters for enterprises still running critical COBOL workloads — developer teams can use these patterns to prioritize modernization with lower upfront engineering effort. Practical case studies indicate measurable reductions in time and cost for common migration tasks.
OpenAI shares reasoning proofs from their models tackling the First Proof math challenge. The results provide insight into the current limits and capabilities of research-grade AI reasoning on expert-level problems.
GGML and llama.cpp — the foundational libraries behind local LLM inference — are joining Hugging Face to ensure their long-term development and maintenance. This brings the most widely-used local inference stack under HF's open-source umbrella, with commitments to continued compatibility and community governance. A significant consolidation in the local AI ecosystem.
IBM Research and UC Berkeley released IT-Bench and MAST — a benchmark suite and failure taxonomy for enterprise AI agents in real IT environments. The research identifies the root causes of enterprise agent failures across categories like tool misuse, context loss, and ambiguity handling. Practical reference for teams debugging why their production agents underperform in enterprise settings.