{"source":"https://vercel.com/changelog","digest":{"id":1,"generated_at":"2026-11-05T04:54:56.700Z","window_start":"2026-10-29T04:54:37.825Z","window_end":"2026-11-05T04:54:37.825Z","item_count":22,"headline":"Vercel hardens agent workflows—and makes cache economics harder to ignore","narrative":"Vercel’s week points toward making agent-driven development safer and easier to operate, rather than simply adding models. [Security Review](https://vercel.com/changelog/vercel-security-review-public-beta) is explicitly in public beta for Pro and Enterprise teams: it checks connected GitHub pushes for vulnerabilities and can hand findings to an agent that proposes a pull request, with approval required before merging. Alongside it, automatic npm blocking moves protection ahead of potentially malicious install scripts. The pattern is security closer to execution, without removing human review of proposed fixes.\n\nThe same emphasis appears in agent infrastructure. The [rewritten fx adapter](https://vercel.com/changelog/fx-harness-adapter-now-separates-host-and-sandbox) keeps orchestration, model requests, and credentials on the host while filesystem and shell operations stay sandboxed. Workflow SDK 5.4.0 makes starts atomic and token-idempotent, closing a duplicate-start race. Daemons gain private service bindings and liveness checks. Meanwhile, `vercel doctor` supplies noninteractive diagnostics and JSON output, and MCP opens connections to clients without prior Vercel allowlisting while retaining user approval. Together, these changes suggest Vercel wants agents to operate more independently, but across clearer isolation and authorization boundaries. AI Gateway’s additional Kimi ZDR providers reinforce that direction: enforced ZDR fails closed when no qualifying provider is available.\n\nThe most immediate builder concern is [Data Cache’s move to Runtime Cache](https://vercel.com/changelog/data-cache-is-moving-to-runtime-cache). This is an exit from beta and a migration, not just a performance announcement: code stays unchanged, storage allocations increase, and previously unbilled reads and writes become metered after migration. Review cache usage before that transition. Elsewhere, efficiency improves without the same trade-off: Sandbox creation with Drives drops from a 1.7-second median to 0.7 seconds automatically, while Flags SDK 1.5.3 and later removes a cold-start network wait and reporting-related function duration. Builders should distinguish automatic improvements from SDK upgrades—and both from billing changes.","themes":[{"name":"Security at execution boundaries","summary":"Public-beta Security Review, automatic npm blocking, and fail-closed ZDR routing put safeguards closer to code changes, builds, and inference."},{"name":"Operational infrastructure for agents","summary":"Host–sandbox separation, atomic workflow starts, daemon health checks, and noninteractive diagnostics make agent workflows more controllable and resilient."},{"name":"Efficiency with explicit economics","summary":"Sandbox and Flags performance improve while the Data Cache migration introduces usage charges that builders should evaluate before switching."}],"model":"cloudflare/openai/gpt-6.1-sol"},"lastRun":{"id":3,"started_at":"2026-11-05T04:53:11.348Z","finished_at":"2026-11-05T04:54:56.813Z","fetched":80,"new_items":0,"status":"ok","error":null},"items":[{"id":"https://vercel.com/changelog/atomic-start-hooks-in-the-workflow-sdk","title":"Workflow SDK adds idempotent starts","url":"https://vercel.com/changelog/atomic-start-hooks-in-the-workflow-sdk","posted_at":"2026-11-04T18:55:00.000Z","summary":"Workflow SDK 5.4.0 adds atomic run admission and hook claiming, using domain tokens to prevent duplicate starts and buffer early hook messages without the previous lookup-then-start race.","theme":"agents","significance":2,"fetched_at":"2026-11-05T04:31:24.800Z","services":[]},{"id":"https://vercel.com/changelog/kimi-k3-1-pro-now-has-more-zdr-providers-on-ai-gateway","title":"Kimi K3.1 Pro now has more ZDR providers on AI Gateway","url":"https://vercel.com/changelog/kimi-k3-1-pro-now-has-more-zdr-providers-on-ai-gateway","posted_at":"2026-11-04T17:00:00.000Z","summary":"AI Gateway now offers Kimi K3.1 Pro and Highspeed through additional Zero Data Retention providers, with request-level or team-wide enforcement that rejects requests when no qualifying provider is available.","theme":"ai-models","significance":1,"fetched_at":"2026-11-05T04:31:24.800Z","services":[]},{"id":"https://vercel.com/changelog/vercel-security-review-public-beta","title":"Vercel Security Review now in public beta","url":"https://vercel.com/changelog/vercel-security-review-public-beta","posted_at":"2026-11-03T20:00:00.000Z","summary":"Vercel Security Review enters public beta for Pro and Enterprise teams, checking connected GitHub pushes for vulnerabilities and offering suggested fixes or agent-generated pull requests that require approval to merge.","theme":"security","significance":3,"fetched_at":"2026-11-05T04:31:24.800Z","services":[]},{"id":"https://vercel.com/changelog/vercel-ci-no-longer-requires-a-linked-vercel-project","title":"Vercel CI no longer requires a linked Vercel project","url":"https://vercel.com/changelog/vercel-ci-no-longer-requires-a-linked-vercel-project","posted_at":"2026-11-03T18:00:00.000Z","summary":"Vercel CI can now run tasks on private GitHub repositories without a linked Vercel project, removing the need to create an empty project solely to connect a repository.","theme":"developer-tools","significance":2,"fetched_at":"2026-11-05T04:31:24.800Z","services":[]},{"id":"https://vercel.com/changelog/tencent-hy4-now-available-on-ai-gateway","title":"Tencent Hy4 now available on AI Gateway","url":"https://vercel.com/changelog/tencent-hy4-now-available-on-ai-gateway","posted_at":"2026-11-03T00:00:00.000Z","summary":"Tencent Hy4 is now available on AI Gateway with a 1M-token context window, vision, reasoning, tool calling, structured outputs, and computer use, at 50% off through December 31, 2026.","theme":"ai-models","significance":2,"fetched_at":"2026-11-05T04:31:24.800Z","services":[]},{"id":"https://vercel.com/changelog/vercel-cli-now-diagnoses-setup-problems-with-vercel-doctor","title":"Diagnose Vercel CLI setup problems with vercel doctor","url":"https://vercel.com/changelog/vercel-cli-now-diagnoses-setup-problems-with-vercel-doctor","posted_at":"2026-11-03T00:00:00.000Z","summary":"The new vercel doctor command diagnoses CLI installation, authentication, team, project, and configuration issues without prompting or changing setup, providing suggested fixes and optional JSON output for coding agents.","theme":"developer-tools","significance":2,"fetched_at":"2026-11-05T04:31:24.800Z","services":[]},{"id":"https://vercel.com/changelog/projects-can-now-be-configured-with-vercel-toml","title":"Projects can now be configured with vercel.toml","url":"https://vercel.com/changelog/projects-can-now-be-configured-with-vercel-toml","posted_at":"2026-11-02T21:00:00.000Z","summary":"All Vercel projects can now use vercel.toml with the same configuration properties as vercel.json, adding comments and a flatter structure while requiring a single project configuration file.","theme":"developer-tools","significance":1,"fetched_at":"2026-11-05T04:31:24.800Z","services":[]},{"id":"https://vercel.com/changelog/vercel-mcp-now-supports-client-id-metadata-documents","title":"Vercel MCP now supports Client ID Metadata Documents","url":"https://vercel.com/changelog/vercel-mcp-now-supports-client-id-metadata-documents","posted_at":"2026-11-02T20:09:00.000Z","summary":"Vercel MCP now accepts clients using HTTPS-hosted Client ID Metadata Documents without prior Vercel allowlisting, while retaining user approval and showing the metadata hostname rather than a verified badge.","theme":"ai-tools","significance":2,"fetched_at":"2026-11-05T04:31:24.800Z","services":[]},{"id":"https://vercel.com/changelog/monitor-vercel-deployments-in-chatgpt","title":"Vercel MCP now renders deployments as an interactive card","url":"https://vercel.com/changelog/monitor-vercel-deployments-in-chatgpt","posted_at":"2026-11-02T17:05:00.000Z","summary":"Vercel MCP now renders interactive deployment cards in ChatGPT and other MCP Apps clients, exposing live status, logs, and commit details with text fallback for clients without app support.","theme":"ai-tools","significance":2,"fetched_at":"2026-11-05T04:31:24.800Z","services":[]},{"id":"https://vercel.com/changelog/flags-explorer-redesigned-for-easier-override-sharing","title":"Flags Explorer redesigned for easier override sharing","url":"https://vercel.com/changelog/flags-explorer-redesigned-for-easier-override-sharing","posted_at":"2026-11-02T16:58:00.000Z","summary":"Flags Explorer now groups overrides, branch recommendations, and related actions together, with a dialog recommending branch overrides to teammates visiting a deployment.","theme":"developer-tools","significance":1,"fetched_at":"2026-11-05T04:31:24.800Z","services":[]},{"id":"https://vercel.com/changelog/vercel-sandbox-with-drives-now-starts-up-to-2-4x-faster","title":"Vercel Sandbox with Drives now starts up to 2.4× faster","url":"https://vercel.com/changelog/vercel-sandbox-with-drives-now-starts-up-to-2-4x-faster","posted_at":"2026-11-02T16:39:00.000Z","summary":"Vercel Sandboxes using Drives now start up to 2.4 times faster, reducing median creation time from 1.7 to 0.7 seconds automatically without SDK changes, configuration changes, or pricing changes.","theme":"infrastructure","significance":2,"fetched_at":"2026-11-05T04:31:24.800Z","services":[]},{"id":"https://vercel.com/changelog/evaluate-vercel-flags-from-another-project-without-sharing-sdk-keys","title":"Evaluate Vercel Flags from another project without sharing SDK keys.","url":"https://vercel.com/changelog/evaluate-vercel-flags-from-another-project-without-sharing-sdk-keys","posted_at":"2026-11-02T16:00:00.000Z","summary":"Vercel Flags Project Connections now provide revocable, read-only access to another project's flags within the same team using deployment identity instead of shared SDK keys, with the consumer deployment determining the environment.","theme":"developer-tools","significance":2,"fetched_at":"2026-11-05T04:31:24.800Z","services":[]},{"id":"https://vercel.com/changelog/longcat-2-5-now-available-on-ai-gateway","title":"LongCat 2.5 now available on AI Gateway","url":"https://vercel.com/changelog/longcat-2-5-now-available-on-ai-gateway","posted_at":"2026-11-02T00:00:00.000Z","summary":"Meituan's LongCat 2.5 is now available on AI Gateway with text and image inputs, a 1M-token context window, and coding and multi-step tool use for complex agent workflows.","theme":"ai-models","significance":2,"fetched_at":"2026-11-05T04:31:24.800Z","services":[]},{"id":"https://vercel.com/changelog/blocking-malicious-npm-packages-before-and-during-installation","title":"Blocking malicious npm packages before and during installation","url":"https://vercel.com/changelog/blocking-malicious-npm-packages-before-and-during-installation","posted_at":"2026-11-02T00:00:00.000Z","summary":"Vercel now automatically checks supported lockfiles and blocks affected public npm downloads during new builds on all plans, failing deployments containing blocked package versions to help prevent malicious install scripts from running.","theme":"security","significance":2,"fetched_at":"2026-11-05T04:31:24.800Z","services":[]},{"id":"https://vercel.com/changelog/data-cache-is-moving-to-runtime-cache","title":"Data Cache is moving to Runtime Cache","url":"https://vercel.com/changelog/data-cache-is-moving-to-runtime-cache","posted_at":"2026-10-30T04:00:00.000Z","summary":"Data Cache is exiting beta and migrating to Runtime Cache without code changes, increasing Pro and Enterprise storage allocations while introducing usage-based billing for previously unbilled reads and writes after migration.","theme":"infrastructure","significance":3,"fetched_at":"2026-11-05T04:31:24.800Z","services":[]},{"id":"https://vercel.com/changelog/vercel-daemons-can-now-call-services-with-bindings","title":"Vercel Daemons can now call services with bindings","url":"https://vercel.com/changelog/vercel-daemons-can-now-call-services-with-bindings","posted_at":"2026-10-30T00:00:00.000Z","summary":"Vercel Daemons can now call private services through service bindings that handle routing, authentication, and TLS while keeping each daemon instance connected to its originating deployment across promotions and rollbacks.","theme":"infrastructure","significance":2,"fetched_at":"2026-11-05T04:31:24.800Z","services":[]},{"id":"https://vercel.com/changelog/ling-3-1-flash-1019-now-available-on-ai-gateway-for-free","title":"Ling 3.1 Flash 1019 now available on AI Gateway for free","url":"https://vercel.com/changelog/ling-3-1-flash-1019-now-available-on-ai-gateway-for-free","posted_at":"2026-10-30T00:00:00.000Z","summary":"InclusionAI's Ling 3.1 Flash 1019 is now available on AI Gateway with a 1M-token context window for coding, reasoning, and tool-using agents, with free requests until published-rate billing begins November 3, 2026.","theme":"ai-models","significance":2,"fetched_at":"2026-11-05T04:31:24.800Z","services":[]},{"id":"https://vercel.com/changelog/add-liveness-checks-to-your-daemons","title":"Add liveness checks to your Daemons","url":"https://vercel.com/changelog/add-liveness-checks-to-your-daemons","posted_at":"2026-10-30T00:00:00.000Z","summary":"Vercel Daemons now support configurable HTTP liveness probes that periodically check a health endpoint and restart the container when it fails, detecting processes that remain running but unhealthy.","theme":"infrastructure","significance":2,"fetched_at":"2026-11-05T04:31:24.800Z","services":[]},{"id":"https://vercel.com/changelog/strix-now-supports-vercel-ai-gateway","title":"Strix now supports Vercel AI Gateway","url":"https://vercel.com/changelog/strix-now-supports-vercel-ai-gateway","posted_at":"2026-10-29T18:44:00.000Z","summary":"The open-source AI penetration testing tool Strix now supports AI Gateway's 400-plus models through a single API, consolidating provider access and usage and cost tracking.","theme":"ai-tools","significance":2,"fetched_at":"2026-11-05T04:31:24.800Z","services":[]},{"id":"https://vercel.com/changelog/improved-ip-geolocation-data","title":"IP geolocation now covers more visitors at higher resolution","url":"https://vercel.com/changelog/improved-ip-geolocation-data","posted_at":"2026-10-29T17:00:00.000Z","summary":"Vercel has automatically improved IP geolocation granularity and refresh frequency across all projects, providing city-level data for nearly all requests and postal codes for 90% of traffic while potentially changing geo-dependent access decisions.","theme":"infrastructure","significance":2,"fetched_at":"2026-11-05T04:31:24.800Z","services":[]},{"id":"https://vercel.com/changelog/vercel-flags-starts-7x-faster-and-no-longer-extends-function-duration","title":"Vercel Flags starts 7x faster and no longer extends function duration","url":"https://vercel.com/changelog/vercel-flags-starts-7x-faster-and-no-longer-extends-function-duration","posted_at":"2026-10-29T06:00:00.000Z","summary":"Vercel Flags SDK 1.5.3 and later removes a cold-start network wait and shifts usage reporting to the runtime, reducing internal-test p75 initialization from over 70 ms to 10 ms and eliminating reporting-related function duration.","theme":"developer-tools","significance":2,"fetched_at":"2026-11-05T04:31:24.800Z","services":[]},{"id":"https://vercel.com/changelog/fx-harness-adapter-now-separates-host-and-sandbox","title":"The fx adapter now runs the harness on the host","url":"https://vercel.com/changelog/fx-harness-adapter-now-separates-host-and-sandbox","posted_at":"2026-10-29T06:00:00.000Z","summary":"The AI SDK's rewritten fx harness adapter runs orchestration and model requests on the host while isolating filesystem and shell operations in the sandbox, keeping model credentials outside it and adding tool filtering, mid-turn steering, structured output, image results, and resumable sessions.","theme":"agents","significance":2,"fetched_at":"2026-11-05T04:31:24.800Z","services":[]},{"id":"https://vercel.com/changelog/node-js-26-lts-is-now-generally-available-on-vercel","title":"Node.js 26 LTS is now Generally Available on Vercel","url":"https://vercel.com/changelog/node-js-26-lts-is-now-generally-available-on-vercel","posted_at":"2026-10-29T00:00:00.000Z","summary":"Node.js 26 LTS is generally available for Vercel builds and Functions and defaults for new projects, bringing HTTP/2-enabled fetch, Temporal, and V8 14.6 while automatically upgrading existing projects whose engine ranges permit 26 on their next deployment.","theme":"infrastructure","significance":3,"fetched_at":"2026-11-05T04:31:24.800Z","services":[]},{"id":"https://vercel.com/changelog/vercel-connect-adds-webhook-triggers-for-more-providers","title":"Vercel Connect adds webhook triggers for more providers","url":"https://vercel.com/changelog/vercel-connect-adds-webhook-triggers-for-more-providers","posted_at":"2026-10-29T00:00:00.000Z","summary":"Vercel Connect adds webhook triggers for Resend, X, Twilio, Notion, and Telegram, verifying requests and handling provider challenges before forwarding events to project endpoints.","theme":"developer-tools","significance":2,"fetched_at":"2026-11-05T04:31:24.800Z","services":[]},{"id":"https://vercel.com/changelog/block-ai-crawlers-and-assistants-independently-with-managed-rules","title":"Control AI crawlers and assistants separately in Vercel Firewall","url":"https://vercel.com/changelog/block-ai-crawlers-and-assistants-independently-with-managed-rules","posted_at":"2026-10-29T00:00:00.000Z","summary":"Vercel Firewall now lets all plans independently log or block AI training and indexing crawlers versus user-triggered AI assistants, allowing teams to restrict collection without blocking assistant retrieval.","theme":"security","significance":2,"fetched_at":"2026-11-05T04:31:24.800Z","services":[]},{"id":"https://vercel.com/changelog/vercel-plugin-is-now-available-in-antigravity","title":"Vercel Plugin is now available in Antigravity","url":"https://vercel.com/changelog/vercel-plugin-is-now-available-in-antigravity","posted_at":"2026-10-29T00:00:00.000Z","summary":"The Vercel Plugin is now available in Google Antigravity's marketplace, giving agents API guidance and recommended patterns through skills for products including AI SDK, Vercel Functions, and Workflow.","theme":"ai-tools","significance":1,"fetched_at":"2026-11-05T04:31:24.800Z","services":[]},{"id":"https://vercel.com/changelog/mistral-large-4-updated-weights-and-lower-pricing-on-ai-gateway","title":"Mistral Large 4 updated weights and lower pricing on AI Gateway","url":"https://vercel.com/changelog/mistral-large-4-updated-weights-and-lower-pricing-on-ai-gateway","posted_at":"2026-10-29T00:00:00.000Z","summary":"AI Gateway's existing mistral/mistral-large-4 ID now uses updated weights and lower input, cached-input, and output pricing, with Mistral reporting improvements in agentic coding, defensive cybersecurity, and prompt-injection resistance.","theme":"ai-models","significance":2,"fetched_at":"2026-11-05T04:31:24.800Z","services":[]},{"id":"https://vercel.com/changelog/vercel-ci-is-now-in-beta","title":"Vercel CI is now in beta","url":"https://vercel.com/changelog/vercel-ci-is-now-in-beta","posted_at":"2026-10-28T19:00:00.000Z","summary":"Vercel CI enters beta for Turborepos, running affected CI-tagged tasks on each push with remote task caching and dependency reuse, with Vercel reporting 2–6 times shorter push-to-pass waits in its internal repositories.","theme":"developer-tools","significance":3,"fetched_at":"2026-11-05T04:31:24.800Z","services":[]},{"id":"https://vercel.com/changelog/box-mount-now-runs-in-vercel-sandbox","title":"Box Mount now runs in Vercel Sandbox","url":"https://vercel.com/changelog/box-mount-now-runs-in-vercel-sandbox","posted_at":"2026-10-28T18:00:00.000Z","summary":"Box Mount, in private preview, now synchronizes Box folders with Vercel Sandbox directories so agents can read and write files through standard filesystem operations while separate sandbox users isolate Box credentials from the agent workspace.","theme":"agents","significance":2,"fetched_at":"2026-11-05T04:31:24.800Z","services":[]},{"id":"https://vercel.com/changelog/vercel-container-registry-now-supports-2-gb-compressed-image-layers-new","title":"Vercel Container Registry now supports 2 GB compressed image layers","url":"https://vercel.com/changelog/vercel-container-registry-now-supports-2-gb-compressed-image-layers-new","posted_at":"2026-10-28T09:32:00.000Z","summary":"Vercel Container Registry increases its compressed per-layer limit from 500 MB to 2 GB while retaining the 15 GB total image limit, allowing larger toolchains and dependency sets without splitting layers.","theme":"infrastructure","significance":1,"fetched_at":"2026-11-05T04:31:24.800Z","services":[]},{"id":"https://vercel.com/changelog/vercel-agent-can-now-validate-alerts-before-paging-you","title":"Vercel Agent can now validate alerts before paging","url":"https://vercel.com/changelog/vercel-agent-can-now-validate-alerts-before-paging-you","posted_at":"2026-10-28T06:00:00.000Z","summary":"Vercel Agent can now investigate monitor alerts using logs and metrics and forward only those it classifies as Critical to selected paging webhooks, available on Pro and Enterprise with Observability Plus.","theme":"observability","significance":2,"fetched_at":"2026-11-05T04:31:24.800Z","services":[]},{"id":"https://vercel.com/changelog/virtual-models-now-available-on-ai-gateway","title":"Virtual models now available on AI Gateway","url":"https://vercel.com/changelog/virtual-models-now-available-on-ai-gateway","posted_at":"2026-10-28T00:00:00.000Z","summary":"AI Gateway now offers virtual models with custom IDs that centrally configure underlying models, inference settings, routing, compliance requirements, and reporting tags without requiring application redeployments or agent reconfiguration.","theme":"ai-tools","significance":3,"fetched_at":"2026-11-05T04:31:24.800Z","services":[]},{"id":"https://vercel.com/changelog/run-paseo-agents-on-vercel-sandbox","title":"Run Paseo agents on Vercel Sandbox","url":"https://vercel.com/changelog/run-paseo-agents-on-vercel-sandbox","posted_at":"2026-10-28T00:00:00.000Z","summary":"Paseo's Vercel Sandbox plugin now runs supported coding agents in cloud sessions controllable from a laptop or phone, preserving files and conversation history when stopped while keeping the AI Gateway key outside the sandbox.","theme":"agents","significance":2,"fetched_at":"2026-11-05T04:31:24.800Z","services":[]},{"id":"https://vercel.com/changelog/particle-1-is-now-available-on-ai-gateway","title":"Particle-1 is now available on AI Gateway","url":"https://vercel.com/changelog/particle-1-is-now-available-on-ai-gateway","posted_at":"2026-10-28T00:00:00.000Z","summary":"Particle-1 is now available on AI Gateway with a 1M-token context window and post-training on DeepSeek V4.1 Flash, targeting software engineering tasks with fewer tokens according to Particle AI.","theme":"ai-models","significance":2,"fetched_at":"2026-11-05T04:31:24.800Z","services":[]},{"id":"https://vercel.com/changelog/agents-can-now-enable-web-analytics-and-observability-plus-with-vercel-cli","title":"Agents can now enable Web Analytics and Observability Plus with Vercel CLI","url":"https://vercel.com/changelog/agents-can-now-enable-web-analytics-and-observability-plus-with-vercel-cli","posted_at":"2026-10-28T00:00:00.000Z","summary":"Vercel CLI 64.2.0 and later can enable Web Analytics and Observability Plus without an interactive terminal, allowing agents to configure linked projects while still requiring an existing team subscription for Observability Plus.","theme":"developer-tools","significance":1,"fetched_at":"2026-11-05T04:31:24.800Z","services":[]},{"id":"https://vercel.com/changelog/fx-is-now-available-in-t3-code","title":"fx is now available in T3 Code","url":"https://vercel.com/changelog/fx-is-now-available-in-t3-code","posted_at":"2026-10-28T00:00:00.000Z","summary":"The open-source fx coding agent harness is now available as a built-in T3 Code provider across web, desktop, and mobile, supporting Supervised and Auto modes alongside existing fx models and skills.","theme":"agents","significance":2,"fetched_at":"2026-11-05T04:31:24.800Z","services":[]},{"id":"https://vercel.com/changelog/vercel-plugin-now-available-in-opencode","title":"Vercel Plugin now available in OpenCode","url":"https://vercel.com/changelog/vercel-plugin-now-available-in-opencode","posted_at":"2026-10-28T00:00:00.000Z","summary":"The Vercel Plugin now supports OpenCode 2 with product skills, specialist agents, deployment and environment commands, and an authenticated MCP connection for accessing projects, deployments, and logs.","theme":"ai-tools","significance":2,"fetched_at":"2026-11-05T04:31:24.800Z","services":[]},{"id":"https://vercel.com/changelog/ai-traffic-insights-now-available-in-vercel-observability","title":"AI traffic insights now available in Vercel Observability","url":"https://vercel.com/changelog/ai-traffic-insights-now-available-in-vercel-observability","posted_at":"2026-10-28T00:00:00.000Z","summary":"Vercel Observability now exposes verified AI bot traffic, crawled pages, assistant retrievals, Markdown requests, and errors in the dashboard, with CLI metric queries available through Observability Plus.","theme":"observability","significance":2,"fetched_at":"2026-11-05T04:31:24.800Z","services":[]},{"id":"https://vercel.com/changelog/deploy-a-container-from-a-plain-dockerfile","title":"Deploy a container from a plain Dockerfile","url":"https://vercel.com/changelog/deploy-a-container-from-a-plain-dockerfile","posted_at":"2026-10-27T00:00:00.000Z","summary":"Vercel's beta Container runtime now automatically builds a root Dockerfile or Containerfile when no other framework is detected, with an explicit Container preset available to override framework detection.","theme":"deployments","significance":2,"fetched_at":"2026-11-05T04:31:24.800Z","services":[]},{"id":"https://vercel.com/changelog/vercel-agent-sessions-now-cost-67-percent-less","title":"Vercel Agent sessions now cost 67% less","url":"https://vercel.com/changelog/vercel-agent-sessions-now-cost-67-percent-less","posted_at":"2026-10-27T00:00:00.000Z","summary":"Vercel Agent dashboard and Slack sessions now cost 67% less on average through more efficient context management, prompt caching, and workflow guidance, with no action required and no change to billing rates.","theme":"agents","significance":2,"fetched_at":"2026-11-05T04:31:24.800Z","services":[]},{"id":"https://vercel.com/changelog/harness-agent-image-inputs","title":"HarnessAgent now supports image inputs for all harness adapters","url":"https://vercel.com/changelog/harness-agent-image-inputs","posted_at":"2026-10-27T00:00:00.000Z","summary":"AI SDK HarnessAgent now accepts mixed text and image inputs through the standard AI SDK message format across every available harness adapter, using each harness's underlying vision model.","theme":"agents","significance":2,"fetched_at":"2026-11-05T04:31:24.800Z","services":[]},{"id":"https://vercel.com/changelog/discover-1000-services-on-vercel-connect-from-the-cli","title":"Discover 1000+ services on Vercel Connect from the CLI","url":"https://vercel.com/changelog/discover-1000-services-on-vercel-connect-from-the-cli","posted_at":"2026-10-26T20:00:00.000Z","summary":"Vercel CLI 64.1.0 and later lets developers and agents search more than 1,000 Connect services with category, connection-method, product, and webhook filters plus JSON output for automation.","theme":"developer-tools","significance":2,"fetched_at":"2026-11-05T04:31:24.800Z","services":[]},{"id":"https://vercel.com/changelog/run-scale-ais-agentenv-evals-on-vercel-sandbox","title":"Run Scale AI’s AgentEnv evals on Vercel Sandbox","url":"https://vercel.com/changelog/run-scale-ais-agentenv-evals-on-vercel-sandbox","posted_at":"2026-10-26T19:00:00.000Z","summary":"Scale AI's AgentEnv 0.9.1334 and later now includes a Vercel provider that runs agents and task environments in separate Sandboxes for evaluations with scores and recorded actions.","theme":"agents","significance":2,"fetched_at":"2026-11-05T04:31:24.800Z","services":[]},{"id":"https://vercel.com/changelog/create-custom-monitors-in-vercel-observability","title":"Create custom monitors in Vercel Observability","url":"https://vercel.com/changelog/create-custom-monitors-in-vercel-observability","posted_at":"2026-10-26T06:00:00.000Z","summary":"Vercel Observability now supports custom query-based monitors with threshold or anomaly detection and multiple notification channels, available on Pro and Enterprise with Observability Plus at $0.0035 per GB scanned.","theme":"observability","significance":2,"fetched_at":"2026-11-05T04:31:24.800Z","services":[]},{"id":"https://vercel.com/changelog/vercel-sandbox-commands-now-support-stdin","title":"Vercel Sandbox commands now support stdin","url":"https://vercel.com/changelog/vercel-sandbox-commands-now-support-stdin","posted_at":"2026-10-26T06:00:00.000Z","summary":"Vercel Sandbox's TypeScript and Python SDKs now support sending standard input to running commands, enabling interaction with stdio MCP servers and other command-line programs.","theme":"infrastructure","significance":2,"fetched_at":"2026-11-05T04:31:24.800Z","services":[]},{"id":"https://vercel.com/changelog/customizable-widgets-for-project-cards","title":"Customizable widgets for project cards","url":"https://vercel.com/changelog/customizable-widgets-for-project-cards","posted_at":"2026-10-26T00:00:00.000Z","summary":"Vercel dashboard project cards now support per-account, per-project metric selections across Firewall, Observability, Speed Insights, and Web Analytics, with feature enablement flows accessible directly from the widget menu.","theme":"developer-tools","significance":1,"fetched_at":"2026-11-05T04:31:24.800Z","services":[]},{"id":"https://vercel.com/changelog/gpt-6-1-terra-now-available-on-ai-gateway","title":"GPT-6.1 Terra now available on AI Gateway","url":"https://vercel.com/changelog/gpt-6-1-terra-now-available-on-ai-gateway","posted_at":"2026-10-26T00:00:00.000Z","summary":"OpenAI's GPT-6.1 Terra is now available on AI Gateway with text and image inputs, a context window exceeding 1M tokens, configurable reasoning and tool calling, plus a fast variant for latency-sensitive visual and computer-use tasks.","theme":"ai-models","significance":2,"fetched_at":"2026-11-05T04:31:24.800Z","services":[]},{"id":"https://vercel.com/changelog/faster-pnpm-installs","title":"Faster pnpm installs","url":"https://vercel.com/changelog/faster-pnpm-installs","posted_at":"2026-10-26T00:00:00.000Z","summary":"Vercel now automatically reuses successful pnpm dependency verification across deployments on all plans, with observed average install times 14% lower for Pro deployments with restored build cache.","theme":"deployments","significance":1,"fetched_at":"2026-11-05T04:31:24.800Z","services":[]},{"id":"https://vercel.com/changelog/vercel-cli-adds-spend-management","title":"Vercel CLI adds Spend Management","url":"https://vercel.com/changelog/vercel-cli-adds-spend-management","posted_at":"2026-10-23T21:44:00.000Z","summary":"Vercel CLI now supports viewing and managing team budgets and alert history, with an optional pause flag to halt production deployments at 100% spend rather than merely setting a budget.","theme":"developer-tools","significance":2,"fetched_at":"2026-11-05T04:31:24.800Z","services":[]},{"id":"https://vercel.com/changelog/send-feedback-from-vercel-mcp","title":"Send feedback from the Vercel MCP","url":"https://vercel.com/changelog/send-feedback-from-vercel-mcp","posted_at":"2026-10-23T20:53:00.000Z","summary":"Vercel MCP adds an authenticated submit_feedback tool that lets agents send user-requested bug reports, friction reports, and feature requests with structured task context.","theme":"ai-tools","significance":1,"fetched_at":"2026-11-05T04:31:24.800Z","services":[]}]}