Cloudflare Daily

A daily briefing on Cloudflare: product launches, the developer changelog, and what the team is shipping.

Cadence: Daily
Length: 2 minutes

Subscribe, Combine, Customize

Subscribe to this podcast
?Receive all episodes to this podcast in the apps below or anywhere that supports RSS.
Combine these episodes into your pod
?All episodes from this podcast will be fed into your own.
Sign up to add to your own podcast
Customize this pod with your own sources
?Use this if you want a brand new podcast with its own episodes using different sources.
Sign up to customize this pod

Sources

Episodes

Cloudflare Daily August 14: Cloudflare Gateway Adds npm and PyPI Package Security Controls
Created: August 14th, 2026 - 04:45 PT
Script

Here is today's Cloudflare Daily for Friday August 14th. Yesterday, Cloudflare Gateway added package registry security, giving organizations more control over a frequently overlooked supply-chain risk: software dependencies downloaded by developers and CI/CD systems. Gateway can now identify package downloads by protocol, extracting details such as ecosystem, package name, version, namespace, and Package URL. Administrators can then write allow or block policies for npm, PyPI, RubyGems, Cargo, Go, Maven, and NuGet packages. Because detection uses the registry protocol rather than a particular hostname, the controls can also work with internal mirrors and proxies such as Artifactory or Nexus. TLS decryption is required. [1]

Also yesterday, Cloudflare Artifacts introduced data-localization support. When creating an Artifacts namespace, teams can now choose the European Union or United States as the sole jurisdiction for repository data storage and processing. Every repository in that namespace inherits the setting. The choice cannot be changed later, so organizations with compliance, sovereignty, or customer-contract requirements should make jurisdiction part of their initial project design rather than treating it as a later migration task. [2]

A third update improves Cloudflare Realtime SFU DataChannels for low-latency applications. Developers can now choose unordered or partially reliable delivery instead of the default reliable, ordered behavior. That is especially useful for game state, sensor readings, chat presence, and control events, where a newer update may be more valuable than a delayed older one. Teams can set no retries, limited retransmissions, or a time limit on delivery, tailoring transport behavior to how quickly information becomes stale. [3]

The broader trend is more precise control at infrastructure boundaries. Cloudflare is giving security teams policy-level visibility into what enters development environments, compliance teams clearer control over where data lives, and application developers more choice over how realtime data behaves. The common lesson is that defaults remain important, but modern systems increasingly need controls that reflect the specific risk, location, and lifetime of each workload. [4]

Thank you for listening to Cloudflare Daily from The Daily FM. See you tomorrow!

Source Evidence
  1. Changelogs | Cloudflare Docs
    ...to Data localization.Aug 13, 2026Detect and control software package downloads with package registry securityCloudflare Gateway can now detect software package downloads and give you policy control over supply chain traffic. When a developer or CI/CD pipeline downloads a package through Gateway, the proxy identifies the registry protocol from the request URL and extracts the package ecosystem, name, version, and namespace. You can then write HTTP policies using pkg.* selectors to allow or block package downloads.
    
    Gateway detects package downloads for the following ecosystems:
    
    Ecosystem
    Namespace
    
    npm
    Scope (for example, @babel)
    
    PyPI
    --
    
    RubyGems
    --
    
    Cargo
    --
    
    Go
    Module path
    
    Maven
    Group ID
    
    NuGet
    --
    
    In the dashboard, select Package Ecosystem to access the package registry selectors. After selecting a single ecosystem, nested fields for package name, version, and n...
  2. Changelogs | Cloudflare Docs
    ...dates and improvements at Cloudflare.Aug 13, 2026Data localization support for ArtifactsArtifacts now supports jurisdictions, allowing you to select the European Union or the United States as the only location where repo data is stored and processed.
    Select a jurisdiction when you create a namespace. Every repo in that namespace automatically uses the selected jurisdiction.
    curl --request POST \
    "https://api.cloudflare.com/client/v4/accounts/$ACCOUNT_ID/artifacts/namespaces" \
    --header "Authorization: Bearer $CLOUDFLARE_API_TOKEN" \
    --header "Content-Type: application/json" \
    --data '{
    "namespace": "my-eu-namespace",
    "jurisdiction": "eu"
    }'
    Jurisdictions cannot be changed after namespace creation. If you omit the jurisdiction, Artifacts creates an unrestricted namespace.
    For supported jurisdictions and usage details, refer to Data localization.Aug 13, 2026Detect and c...
  3. Changelogs | Cloudflare Docs
    ..., game state, sensor updates, and control events.
    
    Realtime SFU DataChannels now support unordered and partially reliable delivery. DataChannels remain reliable and ordered by default, so existing channels keep their current behavior.
    With ordered delivery, a delayed message can block later messages. For game state or sensor updates, recent data may be more useful than recovering an older message. Unordered delivery lets later messages proceed, while partial reliability limits retransmission attempts or delivery time.
    
    Delivery settings answer two questions: whether newer messages can bypass a delayed message, and when the transport should stop retrying delivery.
    Choose the policy that matches how long your payload remains useful:
    
    Goal
    Settings
    Use when
    
    Reliable, ordered delivery (default)
    Omit ordered, maxRetransmits, and maxPacketLifeTime
    Messages remain useful an...
  4. Changelogs | Cloudflare Docs
    ...13, 2026Control Realtime SFU DataChannel delivery
    Cloudflare Realtime SFU is a WebRTC selective forwarding unit that runs on Cloudflare's global network. It forwards audio, video, and application data between WebRTC clients without requiring you to manage SFU infrastructure or regions.
    DataChannels are WebRTC channels for application messages. A client publishes a named DataChannel to Realtime SFU, and the SFU forwards its messages to every client that subscribes to that channel. Use DataChannels for low-latency payloads such as chat messages, game state, sensor updates, and control events.
    
    Realtime SFU DataChannels now support unordered and partially reliable delivery. DataChannels remain reliable and ordered by default, so existing channels keep their current behavior.
    With ordered delivery, a delayed message can block later messages. For game state or sensor updat...
Sources
Cloudflare Daily August 13: Cloudflare WAF Blocks vBulletin RCE as Status Page Goes Machine-Readable
Created: August 13th, 2026 - 04:45 PT
Script

Here is today's Cloudflare Daily for Thursday August 13th. Earlier this week, Cloudflare rebuilt its public Status page while keeping the same address and preserving existing Status API endpoints. The practical change is that notifications now operate independently of Cloudflare infrastructure, with subscriptions available through email, webhooks, Slack, Discord, and Google Chat. The page also now returns Markdown when requested, and separates incidents from maintenance into dedicated RSS and Atom feeds. That is a small but meaningful improvement for both human operators and AI agents: status information is easier to subscribe to, automate against, and consume without brittle HTML parsing. [1]

A second operational update is Cloudflare’s August 11 WAF release. It adds protection for vBulletin CVE-2026-61511, a vulnerability that could enable remote code execution on affected systems. Cloudflare’s release notes indicate the new managed rule is set to block. Two prior beta detections were also folded into their established rules: version-control information disclosure and a vBulletin invalid-image-format code-injection detection. For administrators, the immediate priority remains the same: patch vulnerable applications, but use WAF telemetry and enforcement as a protective layer while remediation is underway. [2]

Yesterday, Cloudflare co-founder Kenton Varda added more color to his experiments running local models with Cloudflare OS. He found that Laguna XS, a mixture-of-experts model, performed closer to dense models than he expected in general application-building tasks. But it fell short on a more visual whiteboard task, producing poor results when asked to draw a monkey. The informal test reinforces that model architecture and benchmark scores do not fully predict real-world agent performance. [3]

The broader trend is operational fit over headline capability. Status systems are becoming machine-readable by default, security rules are being consolidated into clearer enforcement, and teams testing local AI are learning to evaluate models on their actual workflows. The useful question is increasingly not which model or tool is best in the abstract, but whether it is observable, safe, compatible, and reliable for a specific task.

Thank you for listening to Cloudflare Daily from The Daily FM. See you tomorrow!

Source Evidence
  1. Changelogs | Cloudflare Docs
    ...d monitoring continue to work.
    
    The status page now has its own notification system, delivered independently of Cloudflare infrastructure. You can subscribe by email, webhook, Slack, Discord, or Google Chat.
    The Maintenance Notification and Incident Alerts in Cloudflare Notifications remain supported, and deliver to the destinations already configured on your account.
    
    Every page on the status page returns Markdown when requested with an Accept: text/markdown header, so agents can read the current status without parsing HTML:
    curl -H "Accept: text/markdown" https://www.cloudflarestatus.com/locations
    
    Incidents and maintenance are published as separate feeds, each available in RSS and Atom, so you can subscribe to one without the other:
    https://www.cloudflarestatus.com/api/v3/incidents.rss
    https://www.cloudflarestatus.com/api/v3/incidents.atom
    https://www.cloudflaresta...
  2. Changelogs | Cloudflare Docs
    ...en coverage.
    
    Impact
    Successful exploitation of CVE-2026-61511 may lead to remote code execution on affected vBulletin systems, potentially resulting in unauthorized access, data exposure, service disruption, and broader compromise of the hosting environment. Administrators are strongly encouraged to apply vendor updates and recommended mitigations.
    RulesetRule IDLegacy Rule IDDescriptionPrevious ActionNew ActionCommentsCloudflare Managed RulesetN/AvBulletin - Remote Code Execution - CVE:CVE-2026-61511LogBlockThis is a new detection.Cloudflare Managed RulesetN/AVersion Control - Information Disclosure - BetaLogBlockThis rule is merged into the original rule "Version Control - Information Disclosure" (ID: )Cloudflare Managed RulesetN/AvBulletin - Code Injection - Invalid image format - CVE:CVE-2019-17132 - BetaLogBlockThis rule is merged into the original rule "vBullet...
  3. Recent tweets from @KentonVarda
    Laguna XS, however, is an MoE model whose performance seems closer to a dense model.
    
    The one thing it failed at though is when I told it to draw a monkey on the whiteboard that it has just built. It drew some garbage scribbles, whereas all the dense models could draw
    Posted: 2026-08-12T01:37:22.000Z
    Tweet: https://x.com/KentonVarda/status/2087352713705144507
    
    OK, TIL: Some of these models (Qwen, Gemma 4) come in "dense" versions (all parameters active all the time) and mixture-of-expert versions (small subset of parameters active at once). The MoE versions are a lot faster but a lot dumber (for the same model si...
Sources
Cloudflare Daily August 12: Cloudflare Blocks Phishing Content; Kenton Varda Fixes Ollama Prompt Compatibility
Created: August 12th, 2026 - 04:45 PT
Script

Here is today's Cloudflare Daily for Wednesday August 12th. Today, Cloudflare Email Security adds blocked content rules, giving administrators a direct way to stop messages based on words, phrases, or regular-expression patterns in the subject line, body, or both. This is aimed at fast-moving phishing campaigns and organization-specific threats that may not yet be caught by broader reputation or attachment controls. Admins can test regular expressions against sample text before enabling a rule, helping reduce false positives. Matching messages are marked malicious and blocked before they reach inboxes. The feature is available on Enterprise and Enterprise plus PhishGuard packages, and currently supports blocking only, not alternate actions such as quarantine. [1]

Yesterday, Cloudflare co-founder Kenton Varda shared a practical compatibility fix for Cloudflare OS users running local models through Ollama. Some models were not receiving the platform’s system prompt because Cloudflare OS had supplied it using the developer role, which not every model supports. A fix has now been posted. It is a small but important reminder for teams building model-flexible applications: API-level role conventions are not consistently implemented across local and hosted models, so prompt delivery needs validation rather than assumption. [2]

Varda also published informal results from asking several roughly 30-billion-parameter local models to build a collaborative whiteboard application. Muse Glimmer and Gemma 4 performed best in his tests, while a Qwen 3.6 27-billion-parameter variant outperformed the larger 35-billion version he initially tried. He also noted a tradeoff between dense and mixture-of-experts models: mixture-of-experts architectures can be much faster, but may sacrifice capability at a comparable total model size. [3]

The broader trend is operational realism in AI development. New security controls are becoming more customizable, while local-model experimentation is exposing the less glamorous work of compatibility testing, prompt portability, and capability measurement. As organizations adopt agents, the winning systems will not simply select the biggest or fastest model; they will verify behavior, constrain risky inputs, and keep reliable enforcement outside the model itself. [4]

Thank you for listening to Cloudflare Daily from The Daily FM. See you tomorrow!

Source Evidence
  1. Changelogs | Cloudflare Docs
    ...Create rules using either plaintext matches or regular expressions — useful for blocking targeted phishing campaigns, known-bad phrases, or content patterns unique to your organization.
    Choose the search location for each rule: subject, body, or subject and body.
    Use the built-in regular expression checker to validate your pattern against sample text before saving, so you can confirm the rule matches what you expect and avoid false positives.
    Matching messages are marked with a malicious disposition and prevented from reaching users' inboxes.
    
    Blocked content rules currently only support the block action.
    This feature is available for the following Email security packages:
    
    Enterprise
    Enterprise + PhishGuard
    
    To get started, refer to Blocked content.Aug 11, 2026Hostname routing is now generally available, with a new public IP range for initial resolved IPs
    Hostname...
  2. Recent tweets from @KentonVarda
    ...7269635276124624/photo/1
    
    Fix here: https://t.co/iZr6ejjGd1
    Posted: 2026-08-11T19:21:19.000Z
    Tweet: https://x.com/KentonVarda/status/2087258077732942111
    Links: https://github.com/cloudflare/cloudflare-os/pull/157
    
    Ah this turns out to arguably be a problem in Cloudflare OS! We are sending the system prompt to ollama with the "developer" role, but some models support this and some do not. For the ones that do not, I guess, it just drops the content. 🤦
    
    Several people have reported that the
    Posted: 2026-08-11T18:36:31.000Z
    Tweet: https://x.com/KentonVarda/status/2087246801992638781
    
    Ah the problem is ollama is not delivering the system prompt AT ALL to this model. I thought it must be receiving some of the prompt since it successfully wrote a gadget, but it turns out it actually used a blueprint. It knows about tool calls, but not the system prompt! And then https://t....
  3. Recent tweets from @KentonVarda
    ...ing, Laguna XS seems to perform similarly to Glimmer and Gemma.
    
    https://t.co/DSdUM8r68t
    Posted: 2026-08-11T23:01:26.000Z
    Tweet: https://x.com/KentonVarda/status/2087313471058678055
    Links: https://x.com/dartrode/status/2087291136331153795
    
    Wow OK, I was using Qwen 3.6 35B, but there is a 27B variant and... somehow... it's better?
    
    Still performs a bit worse than Glimmer and Gemma in my tests but it's a lot more similar to them.
    Posted: 2026-08-11T22:29:27.000Z
    Tweet: https://x.com/KentonVarda/status/2087305422482862506
    
    OK I was running the 35B version but based on this reply I tried the 27B version and... actually... it's much better? Still not as good as Gemma 4 or Muse Glimmer in my tests but gets a lot closer.
    
    https://t.co/0z4kS2jOwx
    Posted: 2026-08-11T21:53:50.000Z
    Tweet: https://x.com/KentonVarda/status/2087296459385262146
    Links: https://x.com/SecPerkinsStan/st...
  4. Recent tweets from @CloudflareDev
    ...eDev/status/2085393979479380035
    Links: https://blog.cloudflare.com/webmcp/
    
    Agents don't browse like people; they read, discover, call, and pay on a person's behalf.
    
    We're building an open Agentic Internet on standards anyone can implement, so publishers and agents cooperate instead of collide.
    
    https://t.co/wXNk4stT3L
    Posted: 2026-08-06T15:53:32.000Z
    Tweet: https://x.com/CloudflareDev/status/2085393847174283691
    Links: https://blog.cloudflare.com/the-agentic-internet/
    
    A prompt is not a perimeter. Telling an agent "never touch production" shapes behavior but enforces nothing.
    
    The Agent Access Model puts the boundary outside the model with real controls that run at the point of action, scoped to the task at hand.
    
    https://t.co/rYWWmumiL2
    Posted: 2026-08-05T15:56:33.000Z
    Tweet: https://x.com/CloudflareDev/status/2085032217139646541
    Links: https://blog.cloudflare.com/t...
Sources
Cloudflare Daily August 11: Cloudflare Launches Hostname Routing, Tunnel Live Logs, and Turnstile Spin
Created: August 11th, 2026 - 04:45 PT
Script

Here is today's Cloudflare Daily for Tuesday August 11th. Today, Cloudflare has made hostname routing generally available across Cloudflare Tunnel and Cloudflare Mesh. Rather than maintaining static IP lists and routes, teams can direct traffic by hostname: for example, sending a private internal application through a Tunnel, or steering a public hostname through a dedicated egress node. [1]

The launch also changes the default IPv4 range used for initial resolved IPs to Cloudflare-owned public address space. That matters because Chrome 142 is introducing Local Network Access restrictions that can block background requests to the old carrier-grade NAT address range. Since Chromium-based browsers, including Edge, Brave, and Opera, share this behavior, the change aims to prevent hostname-based Gateway features from failing silently. Organizations using custom ranges or legacy Chrome policy workarounds should review their configurations. [2]

Yesterday, Cloudflare added live Cloudflare Tunnel log streaming directly to the main dashboard. The new Live Logs tab can merge output from multiple cloudflared connectors in high-availability deployments, identify the source host for each event, and filter by log level, event type, or HTTP method. This is a practical operational improvement: network teams can investigate a tunnel issue in real time without switching dashboards or collecting logs separately from every replica. [3]

Also yesterday, Turnstile Spin became generally available. Spin helps developers create a Turnstile widget and, critically, connect server-side Siteverify validation to the backend. It can be launched in the dashboard, through Wrangler, or with prompts for AI coding agents. Cloudflare is emphasizing an important anti-bot implementation detail: a widget that appears in the browser but is not validated by the server does not actually protect the request. Spin can test a token against the protected endpoint and replay it to confirm the backend rejects reuse. [4]

The common thread is reducing fragile configuration at the edge. Hostname routing addresses browser-policy changes before they disrupt traffic, live logs shorten incident response, and Turnstile Spin makes secure integration easier to verify. As more infrastructure work is assisted by agents, the useful standard is not just faster setup, but guardrails and observability that prove the setup works. [5]

Thank you for listening to Cloudflare Daily from The Daily FM. See you tomorrow!

Source Evidence
  1. Changelogs | Cloudflare Docs
    ChangelogNew updates and improvements at Cloudflare.Aug 11, 2026Hostname routing is now generally available, with a new public IP range for initial resolved IPs
    Hostname routing ↗ is now generally available. Instead of managing static IP lists and routes, you can route traffic by hostname across multiple Cloudflare One connectors:
    
    Cloudflare Tunnel: route a private hostname (for example, wiki.internal.local) to a private application behind your tunnel, or a public hostname (for example, bank.example.com) to egress through a specific tunnel and anchor traffic to a dedicated exit node.
    Cloudflare Mesh: attract a private or public hostname's traffic to a Mesh node.
    
    Alongside GA, the default IPv4 range used for initial resolved IPs (also called token IPs) is changing from...
  2. Changelogs | Cloudflare Docs
    ...initial resolved IP range for IPv4 if it conflicts with your existing network.
    Why this is changing: Starting with Chrome 142 ↗, Local Network Access (LNA) restrictions block background requests to CGNAT addresses (100.64.0.0/10), which included the previous initial resolved IP default (100.80.0.0/16). LNA is implemented at the Chromium engine level, so it affects all Chromium-based browsers (for example, Microsoft Edge, Brave, and Opera), not only Google Chrome. This could silently break hostname-based Gateway features for users of these browsers, and required Chrome Enterprise policy workarounds. The new default range is public Cloudflare address space, so it is not affected by this restriction.
    What is affected: Initial resolved IPs are used by several features that associate a DNS query with the network connection that follows it:
    
    Private and public hostname rou...
  3. Changelogs | Cloudflare Docs
    ...he dashboard
    Real-time Tunnel log streaming is now available in the Cloudflare dashboard under Networking > Tunnels. This brings the same live debugging capability previously only available in the Cloudflare One dashboard, including multi-connector aggregated streaming for high-availability deployments.
    In the tunnel detail view, a new Live logs tab lets you:
    
    Stream logs from single or multiple connectors — In highly available deployments with multiple cloudflared replicas, logs from all connectors are merged into a single stream grouped by hostname, making it easy to identify which host machine produced each log entry.
    Filter by log level, event type, and HTTP method — Narrow the stream to only the events you care about (HTTP, TCP, UDP, or cloudflared internal), at any log level.
    
    Go to Tunnels ↗For more information, refer to Monitor tunnels and Tunnel log streams.A...
  4. Changelogs | Cloudflare Docs
    ...e widget and wiring canonical server-side siteverify into your existing backend. Start in the dashboard, with Wrangler, or from your AI coding agent. All three paths create the same widget. You can complete the integration by hand or have your agent embed the widget, wire siteverify, and validate it.
    
    Turnstile setup has two parts: embed the widget in your frontend, then call siteverify from your backend. Without the second part, the widget appears on the page but does not protect the request.
    
    The skill includes insertion snippets for Next.js (App Router and Pages Router), Astro, SvelteKit, Hugo, and vanilla HTML. For other frameworks, the agent proposes a generic pattern and asks you to confirm it first.
    The Turnstile dashboard flags existing widgets with no matching siteverify traffic. Select Fix with Spin to copy a prompt that guides your agent through wiring site...
  5. Changelogs | Cloudflare Docs
    ...r entire stack to a private network.
    Kubernetes StatefulSet — deploy a standalone Mesh node with persistent registration state.
    Kubernetes sidecar — add the Mesh image as a sidecar container in a Pod to connect an application to Cloudflare without application changes.
    CI/CD — pull the image in a pipeline step, join the Mesh, run integration tests against private infrastructure, and tear down. The node disappears when the container exits.
    
    For high availability, run multiple replicas with the same Mesh node token. Cloudflare operates replicas in active-passive mode with automatic failover.
    Go to Mesh ↗For setup steps, runtime configuration, and deployment examples, refer to Run Mesh in Docker / Kubernetes.Aug 7, 2026AS-level connectivity and upstream providers on Cloudflare RadarRadar expands its Routing section ↗ with two widgets on AS pages, such as AS13335 ↗, that d...
Sources
Cloudflare Daily August 10: Cloudflare Adds vBulletin RCE and Source-Code Exposure WAF Detections
Created: August 10th, 2026 - 04:45 PT
Script

Here is today's Cloudflare Daily for Monday August 10th. Today, Cloudflare’s scheduled WAF update adds three detections focused on widely used vBulletin deployments and potential source-code exposure. The first is a new rule for CVE-2026-61511, a remote-code-execution vulnerability in vBulletin. The other two are beta detections: one for version-control information disclosure, and another for a vBulletin code-injection issue involving invalid image formats, tracked as CVE-2019-17132. [1]

All three rules are initially set to log rather than block. That is an important operational detail. Logging lets security teams see whether legitimate traffic would be affected before enforcing a block, reducing the risk that a new managed rule disrupts a production application. Teams using Cloudflare’s managed rules should review WAF event logs, identify whether the detections correspond to vulnerable software or expected behavior, and then decide whether to create a more aggressive custom response. [2]

The version-control rule is especially practical. Accidentally exposed Git, SVN, or similar repository artifacts can reveal application code, credentials, deployment history, or configuration files. Even when a detection begins in beta, it is a useful prompt to check that repositories, backups, and build artifacts are never web-accessible. [3]

This update fits a broader security pattern in Cloudflare’s recent releases: controls are increasingly moving from passive visibility toward targeted enforcement. Last week’s WAF release introduced protections for Microsoft SharePoint remote code execution and Rails file-read and RCE risks, while also changing Cloud-hosted SSRF protection from disabled to blocking. The common theme is prioritizing vulnerabilities that can become stepping stones to broader infrastructure compromise. [4]

The practical takeaway is to treat managed WAF rules as both protection and telemetry. Start with logs, validate your exposure, patch the underlying application where necessary, and use enforcement to reduce the window between vulnerability disclosure and remediation. For agent-driven and automated systems especially, security controls need to observe behavior continuously rather than rely on a one-time configuration check. [5]

Thank you for listening to Cloudflare Daily from The Daily FM. See you tomorrow!

Source Evidence
  1. Changelogs | Cloudflare Docs
    ...are changing the action for this rule from Disabled to BLOCKCloudflare Managed RulesetN/ASSRF - Local - BetaDisabled - This detection has been removed.Aug 4, 2026WAF Release - Scheduled changes for 2026-08-10Announcement DateRelease DateRelease BehaviorLegacy Rule IDRule IDDescriptionComments2026-08-042026-08-10LogN/AvBulletin - Remote Code Execution - CVE:CVE-2026-61511This is a new detection.2026-08-042026-08-10LogN/AVersion Control - Information Disclosure - BetaThis is a beta detection and will replace the action on original detection "Version Control - Information Disclosure" (ID: )2026-08-042026-08-10LogN/AvBulletin - Code Injection - Invalid image format - CVE:CVE-2019-17132 - BetaThis is a beta detection and will replace the action on original detection "vBulletin - Code Injection - Invalid image format - CVE:CVE-2019-17132" (ID: )Aug 4, 2026AI agents can debu...
  2. Recent tweets from @CloudflareDev
    ...re.com/community-program-refresh/
    
    Introducing Radar Researcher: ask a question in plain language and get real, interactive charts of Cloudflare's Internet data back.
    
    Ask "what happened to Iran's Internet during the early-2026 shutdown?" and get an annotated traffic timeline.
    
    https://t.co/2XVWGxDZfX
    Posted: 2026-08-07T17:25:21.000Z
    Tweet: https://x.com/CloudflareDev/status/2085779340760129779
    Links: https://blog.cloudflare.com/introducing-radar-researcher/
    
    The line between human and bot keeps blurring, and a single visit can now be part human, part agent. Here's how we tell good behavior from bad, by earning trust from how a visitor acts over a whole session rather than a one-time check.
    
    https://t.co/CjhrQhTnZH
    Posted: 2026-08-07T17:24:20.000Z
    Tweet: https://x.com/CloudflareDev/status/2085779084546871651
    Links: https://blog.cloudflare.com/good-and-bad-agentic-beha...
  3. Changelogs | Cloudflare Docs
    ...: true,
    storeTools: true,
    });
    Open the Agents tab ↗ in the Cloudflare dashboard to inspect sessions, replay conversations, and view trace waterfalls. For advanced setup, privacy controls, and trace structure, refer to Agent tracing.Aug 4, 2026Build and deploy Artifacts repos on every push
    You can now run your CI/CD pipeline on your Artifacts repo by defining a CI Workflow with the CI SDK ↗, automatically triggered on Artifacts push events.
    This allows you to:
    
    Automatically build and deploy application code stored in Artifacts.
    Run linting, type checking, tests, and other checks on every push.
    Reuse dependencies when the lockfile (i.e. pnpm-lock.yaml) has not changed.
    Stop deployment when a check or build fails.
    Restrict API token access to the deployment step.
    Deploy the output to a Worker or a Workers for Platforms User Worker.
    
    Define your CI steps with @cloudflare...
  4. Changelogs | Cloudflare Docs
    ...to execute arbitrary code using crafted requests.
    CVE-2026-66066: An improper input processing vulnerability in Ruby on Rails Active Storage image variant transformations. This may allow an unauthenticated attacker to perform arbitrary file reads and achieve Remote Code Execution (RCE) using maliciously crafted payload requests.
    Generic Cloud Protections: Added improved detection logic targeting Server-Side Request Forgery (SSRF) in cloud-hosted applications.
    
    RulesetRule IDLegacy Rule IDDescriptionPrevious ActionNew ActionCommentsCloudflare Managed RulesetN/AMicrosoft SharePoint - Remote Code Execution - CVE:CVE-2026-50522LogBlockThis is a new detection.Cloudflare Managed RulesetN/ARails - Arbitrary File Read & RCE - CVE:CVE-2026-66066BlockBlockThis was labeled as File Upload - RCE.Cloudflare Managed RulesetN/ASSRF - LocalDisabled - This detection has been removed.C...
  5. Recent tweets from @CloudflareDev
    ...the early-2026 shutdown?" and get an annotated traffic timeline.
    
    https://t.co/2XVWGxDZfX
    Posted: 2026-08-07T17:25:21.000Z
    Tweet: https://x.com/CloudflareDev/status/2085779340760129779
    Links: https://blog.cloudflare.com/introducing-radar-researcher/
    
    The line between human and bot keeps blurring, and a single visit can now be part human, part agent. Here's how we tell good behavior from bad, by earning trust from how a visitor acts over a whole session rather than a one-time check.
    
    https://t.co/CjhrQhTnZH
    Posted: 2026-08-07T17:24:20.000Z
    Tweet: https://x.com/CloudflareDev/status/2085779084546871651
    Links: https://blog.cloudflare.com/good-and-bad-agentic-behaviors/
    
    We're unifying Workers AI and AI Gateway so you can run any model with a consistent interface and unified billing.
    
    Model-first routing and smart routing are coming soon.
    
    https://t.co/WS0Qv6NrmI
    Posted: 2...
Sources
Cloudflare Daily August 9: Cloudflare Unveils AI Agent Stack With WriteGuard and Workers AI
Created: August 9th, 2026 - 04:45 PT
Script

Here is today's Cloudflare Daily for Sunday August 9th. Yesterday, Cloudflare CEO Matthew Prince posted a brief behind-the-scenes note from the company’s One World Trade Center office, where severe weather added a dramatic backdrop to the close of Cloudflare’s Agents Week. The post was not a product announcement, but it underscored how concentrated this week’s news cycle has been around AI agents, developer tooling, and the infrastructure needed to run them safely. [1]

The most useful takeaway for teams is to look past any individual launch and consider the operating model Cloudflare has been assembling. Recent releases connect model access through a unified Workers AI and AI Gateway path; retrieval through AI Search; browser automation through Kitesurf and WebMCP; and agent controls through identity, tracing, write restrictions, and anomaly detection. These components are designed to work as a system rather than as isolated AI features. [2]

A second theme is the distinction between reading and acting. Search, browsing, and data retrieval can help agents understand a task, but actions such as modifying records, deploying code, or making purchases require tighter controls. Cloudflare’s recent focus on task-scoped access, authenticated identity, WriteGuard, and audit trails reflects a growing industry consensus: prompts alone are not a security boundary. [3]

Finally, efficiency is becoming central to agent architecture. If organizations eventually run many agents per employee or customer, the cost of model calls, browser sessions, data retrieval, and network access can escalate quickly. Cloudflare’s emphasis on lightweight Workers-based execution, centralized billing, caching, and observability points toward a practical rule for builders: instrument usage early, constrain privileges by default, and make every expensive agent action measurable. [4]

Thank you for listening to Cloudflare Daily from The Daily FM. See you tomorrow!

Source Evidence
  1. Cloudflare Blog
    ...ge reports.August 5, 2026The Agent Access ModelThe Agent Access Model proposes a new architecture to secure task-scoped agents using strict identity brokering, continuous mediation, and stateful trust.August 5, 2026How we’re rethinking work at Cloudflare with Cloudflare OSWe built Cloudflare OS to equip our teams to safely rethink how they get work done with AI. The platform brings together the best of our technologies, from our Compute primitives to our Zero Trust suite. This post walks through our journey to give our users the best AI tools available.August 5, 2026Cloudflare OS: an open platform for agents, apps, and workCloudflare OS is an open-source platform that lets everyone in your company build apps, automate work, and safely access internal systems, shaped around what your organization knows and how it operatesAugust 5, 2026WriteGuard: fine-grained controls...
  2. Recent tweets from @CloudflareDev
    ...time check.
    
    https://t.co/CjhrQhTnZH
    Posted: 2026-08-07T17:24:20.000Z
    Tweet: https://x.com/CloudflareDev/status/2085779084546871651
    Links: https://blog.cloudflare.com/good-and-bad-agentic-behaviors/
    
    We're unifying Workers AI and AI Gateway so you can run any model with a consistent interface and unified billing.
    
    Model-first routing and smart routing are coming soon.
    
    https://t.co/WS0Qv6NrmI
    Posted: 2026-08-07T17:23:25.000Z
    Tweet: https://x.com/CloudflareDev/status/2085778853147148454
    Links: https://blog.cloudflare.com/workers-ai-gateway-unification/
    
    Now you can give your agents a search engine for your own data. Point AI Search at your files or a site you own and we handle crawling, embedding, and retrieval.
    
    Get a /search and /mcp endpoint to share with your users, or keep it private to your team.
    
    https://t.co/3O1oJ8rZZM
    Posted: 2026-08-06T15:57:58.000Z
    Tweet: ht...
  3. Recent tweets from @CloudflareDev
    ...Tweet: https://x.com/CloudflareDev/status/2085032217139646541
    Links: https://blog.cloudflare.com/the-agent-access-model/
    
    In four months, our internal AI agents flagged nearly 250,000 potential problems and blocked 16,000 bad merges.
    
    Here's how we gave every Cloudflare employee AI they can actually use, including what worked and where we fumbled.
    
    https://t.co/u3AlNtuml1
    Posted: 2026-08-05T15:55:31.000Z
    Tweet: https://x.com/CloudflareDev/status/2085031958741172266
    Links: https://blog.cloudflare.com/how-we-use-ai-with-cloudflare-os/
    
    Now agents can use your MCP write tools while you stay in control of every change they make.
    
    WriteGuard identifies the agent behind each write, limits which tools it can reach, labels the change in the app, and logs the result. Now in private beta.
    
    https://t.co/bCXIzrYAmO
    Posted: 2026-08-05T15:54:36.000Z
    Tweet: https://x.com/CloudflareD...
  4. Recent tweets from @CloudflareDev
    ....000Z
    Tweet: https://x.com/CloudflareDev/status/2085032217139646541
    Links: https://blog.cloudflare.com/the-agent-access-model/
    
    In four months, our internal AI agents flagged nearly 250,000 potential problems and blocked 16,000 bad merges.
    
    Here's how we gave every Cloudflare employee AI they can actually use, including what worked and where we fumbled.
    
    https://t.co/u3AlNtuml1
    Posted: 2026-08-05T15:55:31.000Z
    Tweet: https://x.com/CloudflareDev/status/2085031958741172266
    Links: https://blog.cloudflare.com/how-we-use-ai-with-cloudflare-os/
    
    Now agents can use your MCP write tools while you stay in control of every change they make.
    
    WriteGuard identifies the agent behind each write, limits which tools it can reach, labels the change in the app, and logs the result. Now in private beta.
    
    https://t.co/bCXIzrYAmO
    Posted: 2026-08-05T15:54:36.000Z
    Tweet: https://x.com/Cloud...
Sources
Cloudflare Daily August 8: Cloudflare Launches Radar Researcher Beta and WebMCP for AI Agents
Created: August 8th, 2026 - 04:45 PT
Script

Here is today's Cloudflare Daily for Saturday August 8th. Yesterday, Cloudflare launched Radar Researcher in beta, an AI assistant for exploring Internet trends using plain-language questions. Available within Cloudflare Radar, it can answer by voice or text, explain traffic charts, and generate interactive visualizations from Radar API data. Cloudflare’s example is investigating Iran’s early-2026 Internet shutdown through an annotated traffic timeline. Radar also added WebMCP support, allowing browser-based AI agents to navigate Radar, search data, scan URLs, and look up domains. The significance is that Internet measurement is becoming more accessible: researchers and operators can ask investigative questions without first learning a specialized analytics interface. [1]

Cloudflare also highlighted a related security challenge: agentic behavior cannot reliably be judged from one request or a one-time bot check. A visitor may now be partly human and partly agent, and Cloudflare says trust should be earned over the course of a session, based on behavior. That points toward a more nuanced model for bot management, where legitimate automated activity needs to be distinguished from abuse without blocking useful agents. [2]

For developers building with generative media, BytePlus’s Seedance 2.5 is now available through Cloudflare AI Gateway. The addition expands the gateway’s model catalog and gives teams another video-generation option alongside the operational controls Cloudflare has been emphasizing: centralized observability, security, routing, and billing. [3]

Finally, Cloudflare is relaunching its community program with another one million dollars for open source. The program will support Ambassadors who grow local Cloudflare communities and Community Engineers maintaining open-source projects. Ambassador applications are now open. [4]

The broader trend is that Cloudflare’s agent strategy is extending beyond models and infrastructure. It now spans how agents discover data, operate browsers, establish trust, access media models, and connect with developer communities. For teams, the practical takeaway is to plan agent systems around visibility and behavioral controls from the start, rather than treating them as add-ons after deployment. [5]

Thank you for listening to Cloudflare Daily from The Daily FM. See you tomorrow!

Source Evidence
  1. Changelogs | Cloudflare Docs
    .... Open Researcher from the header on any Radar page to ask questions by voice or text, receive explanations, and view interactive charts based on Radar API data.
    To ask about a specific chart, select Explain with AI to start a conversation with its underlying data and context.
    You can explore further with suggested follow-up questions, find earlier conversations through searchable history, and share conversations through shareable links.
    Alongside the user-facing Researcher experience, Radar now supports WebMCP, allowing browser-based AI agents to navigate Radar, search data, and use tools such as URL scanning and domain lookup.
    To get started, visit Cloudflare Radar ↗.Aug 6, 2026AI Search makes it easier to build a search engine for your dataAI Search gets you from a data source to a working search endpoint quickly. This release adds what you need to put that endpoin...
  2. Recent tweets from @CloudflareDev
    ...ell good behavior from bad, by earning trust from how a visitor acts over a whole session rather than a one-time check.
    
    https://t.co/CjhrQhTnZH
    Posted: 2026-08-07T17:24:20.000Z
    Tweet: https://x.com/CloudflareDev/status/2085779084546871651
    Links: https://blog.cloudflare.com/good-and-bad-agentic-behaviors/
    
    We're unifying Workers AI and AI Gateway so you can run any model with a consistent interface and unified billing.
    
    Model-first routing and smart routing are coming soon.
    
    https://t.co/WS0Qv6NrmI
    Posted: 2026-08-07T17:23:25.000Z
    Tweet: https://x.com/CloudflareDev/status/2085778853147148454
    Links: https://blog.cloudflare.com/workers-ai-gateway-unification/
    
    Now you can give your agents a search engine for your own data. Point AI Search at your files or a site you own and we handle crawling, embedding, and retrieval.
    
    Get a /search and /mcp endpoint to share with your...
  3. Changelogs | Cloudflare Docs
    ChangelogNew updates and improvements at Cloudflare.Aug 7, 2026Workers AI and AI Gateway unify model access and billing
    Workers AI and AI Gateway now provide a unified path for accessing models and managing inference traffic. Use the same AI binding and REST API to call models hosted on Workers AI or by supported third-party providers, with AI Gateway providing observability, logging, caching, security, and billing controls.
    
    The AI binding supports both Workers AI and third-party models through env.AI.run(). The REST API provides shared /ai/ endpoints with Cloudflare authentication across providers.
    Route a Workers AI request through AI Gateway by specifying a gateway ID. Use default to auto...
  4. Recent tweets from @CloudflareDev
    Seedance 2.5 from @BytePlusGlobal is now available on
    @Cloudflare AI Gateway. 
    
    https://t.co/MlruCENoV3
    Posted: 2026-08-07T20:24:25.000Z
    Tweet: https://x.com/CloudflareDev/status/2085824404559192073
    Links: https://developers.cloudflare.com/ai/models/bytedance/seedance-2.5/
    
    We're relaunching our community program with another $1M for open source, backing Ambassadors who bring Cloudflare to their local communities and Community Engineers who maintain open-source projects.
    
    Ambassador applications are open now.
    
    https://t.co/JNg78szQRT
    Posted: 2026-08-07T17:26:03.000Z
    Tweet: https://x.com/CloudflareDev/status/2085779518539837767
    Links: https://blog.cloudflare.com/commu...
  5. Recent tweets from @CloudflareDev
    ...ta back.
    
    Ask "what happened to Iran's Internet during the early-2026 shutdown?" and get an annotated traffic timeline.
    
    https://t.co/2XVWGxDZfX
    Posted: 2026-08-07T17:25:21.000Z
    Tweet: https://x.com/CloudflareDev/status/2085779340760129779
    Links: https://blog.cloudflare.com/introducing-radar-researcher/
    
    The line between human and bot keeps blurring, and a single visit can now be part human, part agent. Here's how we tell good behavior from bad, by earning trust from how a visitor acts over a whole session rather than a one-time check.
    
    https://t.co/CjhrQhTnZH
    Posted: 2026-08-07T17:24:20.000Z
    Tweet: https://x.com/CloudflareDev/status/2085779084546871651
    Links: https://blog.cloudflare.com/good-and-bad-agentic-behaviors/
    
    We're unifying Workers AI and AI Gateway so you can run any model with a consistent interface and unified billing.
    
    Model-first routing and smart rout...
Sources
Cloudflare Daily August 7: Cloudflare Unifies AI Access, Launches Docker Mesh and Kitesurf Beta
Created: August 7th, 2026 - 04:45 PT
Script

Here is today's Cloudflare Daily for Friday August 7th. Today, Cloudflare unified Workers AI and AI Gateway into one path for model access, observability, and billing. Developers can use the same AI binding or REST API to call Cloudflare-hosted models and supported third-party providers, while AI Gateway adds logging, caching, rate limiting, retries, security controls, and cost analytics. Prepaid AI Gateway credits can now cover Workers AI inference too, including access to selected frontier models without a Workers Paid plan. The practical benefit is simpler multi-model operations: one integration, one credit balance, and clearer controls over where AI spend goes. [1]

Also today, Cloudflare Mesh became available as a Docker container. Teams can run Mesh nodes in Docker Compose, Kubernetes, CI pipelines, or other OCI-compatible environments without installing packages on the host. The image supports both AMD64 and ARM64, includes source NAT to avoid VPC route-table changes, and can be replicated for active-passive failover. That should make it much easier to attach temporary workloads, test environments, and containerized applications to private infrastructure. [2]

Yesterday, Cloudflare expanded AI Search, its managed system for building search and retrieval over a company’s files and websites. New options let developers use custom domains, put Cloudflare Access authentication in front of endpoints, and expose a namespace-wide search, chat-completions, or MCP endpoint across selected indexes. A new crawler mode can discover linked pages beyond incomplete sitemaps. The result is a more production-ready way to give internal or customer-facing agents grounded access to owned data. [3]

Cloudflare also introduced Kitesurf in beta, a stateless browser built for AI agents and run entirely on Workers. Cloudflare says it uses three to seven times less CPU and memory than Chromium for common tasks such as screenshots and HTML extraction. The larger trend is efficiency at agent scale: as organizations run more agents, browsers, model calls, retrieval systems, and private-network connections all need to become cheaper, observable, and governed by default. [4]

Thank you for listening to Cloudflare Daily from The Daily FM. See you tomorrow!

Source Evidence
  1. Changelogs | Cloudflare Docs
    ...ests routed through AI Gateway can be logged and included in analytics for request volume, errors, latency, token usage, and costs. You can also configure controls such as caching, rate limiting, and request retries on the gateway.
    
    You can now use prepaid AI Gateway credits to pay for Workers AI inference. This provides one credit balance for Workers AI and supported third-party model providers. To use credits for Workers AI, set the gateway's Workers AI billing setting to Unified billing. Workers AI requests routed through that gateway deduct from your credit balance in real time.
    Prepaid credits also provide access to the following Workers AI frontier models without requiring the Workers Paid plan. Each frontier Workers AI model has a rate limit of 50 requests per minute per account, per model when billed with AI Gateway credits, compared to 20 requests per minute...
  2. Changelogs | Cloudflare Docs
    ...or Docker Compose, Kubernetes, and any OCI-compatible runtime — no host-level package installation required.
    The image supports amd64 and arm64 architectures and includes built-in source NAT so return traffic routes correctly without VPC route table changes.
    
    Docker Compose — add a cloudflare-mesh service to your compose.yaml and connect your entire stack to a private network.
    Kubernetes StatefulSet — deploy a standalone Mesh node with persistent registration state.
    Kubernetes sidecar — add the Mesh image as a sidecar container in a Pod to connect an application to Cloudflare without application changes.
    CI/CD — pull the image in a pipeline step, join the Mesh, run integration tests against private infrastructure, and tear down. The node disappears when the container exits.
    
    For high availability, run multiple replicas with the same Mesh node token. Cloudflare operate...
  3. Changelogs | Cloudflare Docs
    ...her than to anyone who finds the URL. Agents authenticate with an Access service token, and people who open the endpoint in a browser sign in through your identity provider.
    
    A namespace can expose its own public endpoint with /search, /chat/completions, and /mcp paths that fan out across the instances you choose:
    curl https://ns-<NAMESPACE_ENDPOINT_ID>.search.ai.cloudflare.com/search \
    --header "Content-Type: application/json" \
    --data '{
    "messages": [{ "content": "How do I configure AI Search?", "role": "user" }],
    "ai_search_options": { "instance_ids": ["docs", "support"] }
    }'
    
    Website data sources support a new discover parse type. It starts at the source URL and collects pages from both your sitemaps and the links it finds while crawling:
    curl -X POST "https://api.cloudflare.com/client/v4/accounts/<ACCOUNT_ID>/ai-search/instances" \
    -H "Authorization: Bearer <API_...
  4. Changelogs | Cloudflare Docs
    ...type": "web-crawler",
    "source": "example.com",
    "source_params": {
    "web_crawler": {
    "parse_type": "discover",
    "discover_options": { "source": "links", "limit": 5000, "depth": 3 }
    }
    }
    }'
    To learn more, refer to the AI Search documentation.Aug 6, 2026Introducing Kitesurf, an agent-first browser on Browser RunKitesurf is Cloudflare's new stateless, highly scalable browser that runs entirely on top of Workers and is designed for AI agents. It is available for free while in beta.
    Compared to Chromium, Kitesurf uses 3–7× less CPU and memory for common agentic tasks like screenshots and HTML extraction, so you can run more sessions and scale better for bursty, AI-driven workloads.
    Your existing clients already work. To opt in, add the browser=kitesurf parameter to any Browser Run CDP or Quick Action endpoint:
    curl -X POST 'https://api.cloudflare.com/client/v4/accounts/<ACCOUN...
Sources
Cloudflare Daily August 6: Cloudflare Launches OS, WriteGuard and AI Gateway User Insights
Created: August 6th, 2026 - 04:45 PT
Script

Here is today's Cloudflare Daily for Thursday August 6th. Yesterday, Cloudflare unveiled Cloudflare OS, an open-source platform for giving employees AI agents and workspaces that understand company context and can safely reach internal systems. The platform combines Cloudflare’s compute, Zero Trust, and application-building tools so organizations can create internal apps, automate workflows, and connect agents to approved data and services. Cloudflare says it has already used the approach internally: its AI agents identified nearly 250,000 potential issues and blocked 16,000 problematic code merges over four months. Developers can deploy Cloudflare OS into their own Cloudflare accounts now. [1]

Security is the central companion announcement. Cloudflare’s new Agent Access Model argues that a prompt telling an agent not to touch production is not a meaningful security boundary. Instead, agents should receive task-scoped identities, with access brokered and continuously checked at the point where an action occurs. In practical terms, permissions should be narrow, temporary, and tied to the specific task, rather than inherited broadly from the human who launched the agent. [2]

Cloudflare also put that idea into a product feature with WriteGuard, now in private beta for MCP server portals. WriteGuard identifies the agent making a write request, limits which tools it can use, labels the resulting change in the target application, and records the outcome. The goal is to let agents use powerful write tools without requiring employees to perfectly configure every agent or manually supervise every tool call. [3]

Finally, AI Gateway has added User Insights at no additional cost. The dashboard breaks down AI cost, tokens, requests, model usage, and cache behavior by individual user. More notably, it establishes each user’s normal session-spend baseline and flags unusually expensive activity, which can point to compromised credentials or runaway agents before the bill arrives. [4]

The emerging pattern from Agents Week is clear: Cloudflare is framing enterprise AI less as a chatbot feature and more as an operating environment. The infrastructure priorities are identity, least-privilege access, audit trails, anomaly detection, and company-specific context. Thank you for listening to Cloudflare Daily from The Daily FM. See you tomorrow! [5]

Source Evidence
  1. The Cloudflare Blog
    ...k done with AI. The platform brings together the best of our technologies, from our Compute primitives to our Zero Trust suite. This post walks through our journey to give our users the best AI tools available.August 5, 2026Cloudflare OS: an open platform for agents, apps, and workCloudflare OS is an open-source platform that lets everyone in your company build apps, automate work, and safely access internal systems, shaped around what your organization knows and how it operatesAugust 5, 2026WriteGuard: fine-grained controls for MCP ServersAt Cloudflare, we knew we could not depend on every employee to configure every agent perfectly or watch every tool call. Before expanding write access across our own internal MCP servers, we built WriteGuard. We are now bringing those controls to Cloudflare MCP server portals through a private beta.
    August 4, 2026How Cloudflare enf...
  2. Recent tweets from @CloudflareDev
    A prompt is not a perimeter. Telling an agent "never touch production" shapes behavior but enforces nothing.
    
    The Agent Access Model puts the boundary outside the model with real controls that run at the point of action, scoped to the task at hand.
    
    https://t.co/rYWWmumiL2
    Posted: 2026-08-05T15:56:33.000Z
    Tweet: https://x.com/CloudflareDev/status/2085032217139646541
    Links: https://blog.cloudflare.com/the-agent-access-model/
    
    In four months, our internal AI agents flagged nearly 250,000 potential problems and blocked 16,000 bad merges.
    
    Here's how we gave every Cloudflare employee AI they can actually use, including what worked and where we fumbled.
    
    https://t.co/u3AlNtuml1
    Posted: 2026-08-05T15:55:31.000Z
    Tweet: https://x.com/CloudflareDev/status/2085031958741172266
    Links: https://blog.cloudflare.com/how-we-use-a...
  3. Recent tweets from @CloudflareDev
    ...s://t.co/u3AlNtuml1
    Posted: 2026-08-05T15:55:31.000Z
    Tweet: https://x.com/CloudflareDev/status/2085031958741172266
    Links: https://blog.cloudflare.com/how-we-use-ai-with-cloudflare-os/
    
    Now agents can use your MCP write tools while you stay in control of every change they make.
    
    WriteGuard identifies the agent behind each write, limits which tools it can reach, labels the change in the app, and logs the result. Now in private beta.
    
    https://t.co/bCXIzrYAmO
    Posted: 2026-08-05T15:54:36.000Z
    Tweet: https://x.com/CloudflareDev/status/2085031726745788425
    Links: https://blog.cloudflare.com/mcp-portal-writeguard-private-beta/
    
    Now you can see who is spending on AI, down to the person or agent behind every request.
    
    Identity-aware AI Gateway (open beta) puts a verified user on each request. User Insights (GA, free) baselines each one and flags the anomalies.
    
    https://t.co/W2Tg...
  4. Changelogs | Cloudflare Docs
    ...totals for cost, requests, tokens, and adoption, and lets you drill into an individual user to see their spend, top models and providers, cache hit rate, and more. To attribute usage to individual users, add a user identifier with custom metadata or put your gateway behind Cloudflare Access.
    On the security side, User Insights baselines each user's normal usage from their 95th percentile (p95) session cost over the last 30 days, then flags sessions that exceed both that baseline and an organization-level threshold. A sudden jump above a user's own pattern is often the first sign of a compromised credential or a misbehaving agent, so you can investigate before it shows up on your bill.
    User Insights is available to all AI Gateway customers at no additional cost.Aug 5, 2026Identity-aware controls are now available in AI GatewayAI Gateway now integrates with Cloudflare...
  5. Recent tweets from @CloudflareDev
    ...n.
    
    First up is agent tracing: see every model call, tool execution, and token as a session replay and trace waterfall.
    
    https://t.co/Hcwbxrofos
    Posted: 2026-08-04T18:00:58.000Z
    Tweet: https://x.com/CloudflareDev/status/2084701142576238715
    Links: https://blog.cloudflare.com/agents-on-cloudflare/
    
    AI made writing code the fastest, cheapest step in the software development lifecycle. Everything downstream is now overwhelmed.
    
    Here's how agents should take on the rest of the work, not just the code.
    https://t.co/EZ0vXsIXnZ
    Posted: 2026-08-04T17:54:32.000Z
    Tweet: https://x.com/CloudflareDev/status/2084699523570655650
    Links: https://blog.cloudflare.com/agent-development-lifecycle/
    
    Agents Week is here!
    
    To celebrate, we're offering $100 off Cloudflare Connect San Francisco (October 19-21). Use code CFAGENTS26 at registration - valid through August 9.
    
    https://t.co/cl40B4iY...
Sources
Cloudflare Daily August 5: Cloudflare Adds AI Gateway Identity Controls, Agents Tracing and Codex
Created: August 5th, 2026 - 04:45 PT
Script

Here is today's Cloudflare Daily for Wednesday August 5th. Today’s top update gives AI Gateway identity-aware controls through Cloudflare Access. Teams can now put an AI Gateway endpoint behind Access policies and use a verified user identity throughout gateway logs, analytics, routing, and spend controls. That means organizations can set per-user limits, restrict access to particular gateways, and investigate usage without trusting a client application to supply its own user IDs. For enterprise AI deployments, that is a meaningful move from API-key-level governance toward user-level accountability. [1]

Yesterday, Tuesday, Cloudflare introduced Cloudflare Agents, a central experience for deploying and operating agent applications at scale. Its first major capability is agent tracing: developers can inspect sessions as conversation replays and trace waterfalls, following each agent turn through model calls, tool executions, approvals, token use, and Workers runtime operations. Tracing works with Think and Flue automatically, while direct AI SDK users can instrument calls through a wrapper. Sensitive message and tool-payload recording remains off by default, an important privacy safeguard. [2]

Cloudflare also announced the Cloudflare Codex, a governed collection of engineering standards designed for the AI-assisted development lifecycle. Structured RFCs, owned by domain experts, can be consumed by coding agents and paired with agentic reviews. The aim is to enforce consistency not just in generated code, but in specifications and incident reports as well. It addresses a growing problem: AI makes producing code fast, but review, reliability, and organizational standards can become the bottleneck. [3]

One more practical rollout from yesterday: Artifacts repositories can now trigger Cloudflare CI Workflows on every push. Teams can run isolated install, lint, test, type-check, build, and deployment stages, cache dependencies based on lockfiles, halt deploys when checks fail, and limit API-token access to the deployment step. CloudflareDev says the system also supports retries and a self-healing agent for failed steps. [4]

The trend is clear: Cloudflare’s Agents Week is focusing less on simply creating agents and more on operating them safely. Identity, observability, standards, and automated delivery are becoming core infrastructure for production AI systems. Thank you for listening to Cloudflare Daily from The Daily FM. See you tomorrow! [5]

Source Evidence
  1. Changelogs | Cloudflare Docs
    ...tewayAI Gateway now integrates with Cloudflare Access, giving you two new capabilities:
    
    Protect your gateway endpoint. Put your AI Gateway behind Access so you can set policies that control who is allowed to call a specific gateway's endpoint.
    Identity-aware controls. When traffic reaches AI Gateway through an Access-protected custom domain, AI Gateway can use the authenticated user's Access identity in logs, analytics, routing, and spend controls.
    
    With identity-aware controls, you can set spend limits by authenticated user, control which gateways different users can access, filter logs by user, and build policies without passing user IDs from the client application. AI Gateway adds the verified Access user ID to request metadata as cf.user_id.
    For setup instructions, refer to Cloudflare Access.Aug 4, 2026Agent traces for Think, Flue, and AI SDK instrumented by Agen...
  2. Changelogs | Cloudflare Docs
    ...nt application. AI Gateway adds the verified Access user ID to request metadata as cf.user_id.
    For setup instructions, refer to Cloudflare Access.Aug 4, 2026Agent traces for Think, Flue, and AI SDK instrumented by Agents SDK
    Agent tracing is now available for applications built with the Agents SDK. Traces show each agent turn alongside model calls, tool runs, approvals, token usage, and Workers runtime operations.
    Turn on Workers tracing in your Wrangler configuration:
    {
    "$schema": "./node_modules/wrangler/config-schema.json",
    "observability": {
    "traces": {
    "enabled": true
    }
    }
    }[observability.traces]
    enabled = true
    Think and Flue applications emit agent traces automatically. For direct AI SDK calls, wrap the AI SDK namespace once. wrapAISDK() supports AI SDK v6 and v7. This AI SDK v7 example also supplies the agent identity:
    import * as ai from "ai";
    import { wrapAISD...
  3. The Cloudflare Blog
    August 4, 2026How Cloudflare enforces engineering standards using AIWe created the Cloudflare Codex, a governed body of engineering standards that AI agents consume across the development lifecycle. By pairing structured RFCs with agentic reviews, teams automatically enforce consistency across code, specs, and incident reports.
    August 4, 2026Introducing: Cloudflare AgentsCloudflare Agents brings all of your deployed agent sessions into a single experience, surfacing key information and insights into how your agents perform at scale.August 3, 2026Introducing the Billable Usage API: programmatic cost visibility for CloudflareCloudfl...
  4. Recent tweets from @CloudflareDev
    ...om @bfl_ai is now available on
    @Cloudflare AI Gateway. The leading multimodal video model on the fastest gateway.
    
    https://t.co/o0R4dBebnL
    Posted: 2026-08-04T22:07:51.000Z
    Tweet: https://x.com/CloudflareDev/status/2084763271093993770
    Links: https://developers.cloudflare.com/ai/models/black-forest-labs/flux-3-video/
    
    Run your CI/CD pipeline as a Cloudflare Workflow: install, lint, test, and deploy.
    
    Each step supports built-in retries, plus a self-healing agent that can fix broken steps and push commits.
    
    Request access: https://t.co/1r1Rfw9AHf
    Posted: 2026-08-04T18:36:43.000Z
    Tweet: https://x.com/CloudflareDev/status/2084710136950714679
    Links: https://blog.cloudflare.com/ci-workflows/
    
    Astro used isolated AI subagents in GitHub Actions to reproduce issues and verify fixes before merging, taking its open issue count from 200+ to about 30. 
    
    The code is public: fork wit...
  5. Changelogs | Cloudflare Docs
    ...r, control which gateways different users can access, filter logs by user, and build policies without passing user IDs from the client application. AI Gateway adds the verified Access user ID to request metadata as cf.user_id.
    For setup instructions, refer to Cloudflare Access.Aug 4, 2026Agent traces for Think, Flue, and AI SDK instrumented by Agents SDK
    Agent tracing is now available for applications built with the Agents SDK. Traces show each agent turn alongside model calls, tool runs, approvals, token usage, and Workers runtime operations.
    Turn on Workers tracing in your Wrangler configuration:
    {
    "$schema": "./node_modules/wrangler/config-schema.json",
    "observability": {
    "traces": {
    "enabled": true
    }
    }
    }[observability.traces]
    enabled = true
    Think and Flue applications emit agent traces automatically. For direct AI SDK calls, wrap the AI SDK namespace once. wrapAIS...
Sources

<- Back to library