Skip to content
Cloud Security DeskSearch
Menu

Research catalogue

Publications

187 open-access publications for people responsible for cloud control evidence.

From the desk

187 publications
Technical guideSource-based analysis

Stop pull_request_target workflows from running fork code

GitHub blocks pull_request_target in public repositories by default from November 2, 2026. Find the workflows that run fork code with secrets, split them, pin a checkout that refuses unsafe refs, and allow the rest deliberately.

Workload security · GitHub · By Cloud Security Desk
Technical guideSource-based analysis

Run AI coding agents in GitHub Actions without exposing secrets

Treat the agent job as an untrusted-input processor. Follow the outside text from the trigger that admits it to the credentials it can reach, then split reading from writing so a successful injection reaches nothing worth stealing.

AI systems · GitHub / Anthropic / Google / Microsoft · By Cloud Security Desk
Technical guideSource-based analysis

Move Microsoft Sentinel to the Defender portal without breaking automation

Onboarding a Sentinel workspace to the Defender portal changes incident providers, grouping, fields and automation timing with no documented error. Inventory what depends on them, rehearse safely and cut over before March 31, 2027.

Detection & response · Microsoft Sentinel / Microsoft Defender · By Cloud Security Desk
Technical guideSource-based analysis

Treat a cloud model retirement as a security change

Bedrock and Agent Platform fail closed on a model's retirement date, while Foundry can swap a different model in behind the same deployment name. Read the three calendars, compare the replacement's controls and decide per deployment.

AI systems · Amazon Web Services / Microsoft Azure / Google Cloud · By Cloud Security Desk
Technical guideSource-based analysis

Respond to an exposed AWS access key beyond the quarantine policy

In a published test, AWS attached its quarantine policy 10 seconds after a key leaked, but the policy denies actions without ending access. This playbook covers what it leaves open, which credentials outlive the key, and the order of containment.

Detection & response · Amazon Web Services / GitHub / Google Cloud · By Cloud Security Desk
Technical guideSource-based analysis

Admit MCP servers before agents can call their tools

An MCP server's tool descriptions can change after approval with no new release. Record provenance and execution context, hash the definitions per authorization context, enforce the pin at a gateway or allowlist, and watch for drift.

AI systems · MCP / AWS / Anthropic / GitHub / Microsoft Azure / Google Cloud / OWASP · By Cloud Security Desk
Technical guideSource-based analysis

Roll out configuration changes as carefully as code

Flags, policy rows, generated files and customer settings triggered several global outages between 2024 and 2026. Seven primary incident reports show how each change slipped past its safeguards, and which controls their authors adopted afterward.

Resilience · Amazon Web Services / Microsoft Azure / Google Cloud / Cloudflare / CrowdStrike · By Cloud Security Desk
Technical guideSource-based analysis

Roll out Entra token protection without breaking unsupported clients

Token protection makes Entra ID refuse token requests that a device-bound session does not back. Scope it to supported apps and platforms, triage report-only data by code and platform, and cover what stays unbound.

Identity & access · Microsoft Entra / Microsoft Intune / Microsoft Graph · By Cloud Security Desk
Technical guideSource-based analysis

Plan for incident reporting clocks that start at different moments

GDPR, NIS2, DORA, the SEC, the US banking agencies and CERT-In each start their reporting clock at a different decision. Plan around those triggers, the evidence each first notice needs and the rules still pending.

Detection & response · European Union / U.S. Securities and Exchange Commission / U.S. federal banking agencies / CISA / CERT-In / UK Parliament / Amazon Web Services / Microsoft Azure / Google Cloud · By Cloud Security Desk
Technical guideSource-based analysis

Investigate a SaaS integration breach through its OAuth tokens

A vendor that stores OAuth tokens for your SaaS tenants has been breached. Confirm from your own logs what the tokens did, revoke them on every platform, and search the exposed records for secrets.

Detection & response · Salesforce / Google Workspace / Microsoft Entra ID · By Cloud Security Desk
Technical guideSource-based analysis

Investigate stolen cloud credentials used to run AI models

When stolen keys run foundation models in your account, the evidence is split between CloudTrail, usage logs that may never have been enabled and a bill that lags by a day or more.

Detection & response · Amazon Web Services / Microsoft Azure / Google Cloud · By Cloud Security Desk
Technical guideSource-based analysis

Tier Entra administrator roles by the control they can take

Microsoft's own password reset and sensitive action tables show which Entra roles can take over which accounts. Use them, plus the non-role paths, to set three tiers, assignment rules and the controls that keep lower tiers out.

Identity & access · Microsoft Entra / Microsoft Graph · By Cloud Security Desk
Technical guideSource-based analysis

Find the Kubernetes RBAC permissions that lead to cluster admin

Group RBAC grants by what they yield, not by verb. Then close the nodes/proxy exec path with v1.36 fine-grained kubelet permissions, narrow impersonation, and prove each removal with a negative test.

Identity & access · Kubernetes / Amazon Web Services / Google Cloud / Microsoft Azure · By Cloud Security Desk
Technical guideSource-based analysis

Know what service mesh mTLS proves about the caller

A mesh handshake proves that a peer held a key for a namespace and service account. It says nothing about the user or the operation, and permissive defaults, probes and waypoints open paths around it.

Workload security · Istio / Linkerd / Kubernetes / Google Cloud / Microsoft Azure / Amazon Web Services · By Cloud Security Desk
Technical guideSource-based analysis

Find the AWS global control planes in your recovery path

Thirteen AWS control planes run in us-east-1 or us-west-2, and ordinary calls from other Regions still reach them. Audit runbooks, SDK settings and the last exercise, then swap each dependency for a data-plane operation.

Resilience · Amazon Web Services · By Cloud Security Desk
Technical guideSource-based analysis

Lock down self-hosted model servers before exposing them

Built-in credentials in vLLM, SGLang and Triton cover less than their names suggest, Ollama's local API has none, and vLLM and SGLang leave their internal planes unauthenticated. Here is what stays open and what must sit in front.

AI systems · vLLM / SGLang / Ollama / NVIDIA Triton Inference Server / Kubernetes / nginx · By Cloud Security Desk
Technical guideSource-based analysis

Tune managed WAF rules without opening a bypass

Fix a managed WAF false positive by changing one rule for one route, then decide what happens to the request body bytes AWS WAF, Azure WAF and Cloud Armor never inspect.

Workload security · Amazon Web Services / Microsoft Azure / Google Cloud · By Cloud Security Desk
Technical guideSource-based analysis

Plan for an outage at your CDN or edge provider

When the CDN, WAF and edge provider fails, traffic can move to a regional WAF, a second CDN or the origin, and each keeps less protection. Decide early whether a second path is worth securing.

Resilience · Microsoft Azure / Cloudflare / Amazon Web Services / Google Cloud · By Cloud Security Desk
Technical guideSource-based analysis

Find the Entra changes that Backup and Recovery cannot roll back

Microsoft Entra Backup and Recovery rolls listed properties back to a daily backup kept for seven days. Credentials, role assignments, owners and user consent stay in place, and an unscoped recovery can undo your own containment.

Resilience · Microsoft Entra / Microsoft Graph · By Cloud Security Desk
Technical guideSource-based analysis

Verify EC2 containment beyond a security group change

Swapping a compromised EC2 instance into an isolation security group stops new connections, not tracked ones. Learn which flows survive, how long idle timeouts let them run, and how to prove the cut happened.

Detection & response · Amazon Web Services · By Cloud Security Desk
Technical guideSource-based analysis

Use DNS egress filtering without mistaking it for a firewall

Resolver-layer filters in AWS, Azure and Google Cloud decide which names a workload can resolve, not where it connects. What each provider says its DNS control misses, and the connection control to pair with it.

Workload security · Amazon Web Services / Microsoft Azure / Google Cloud · By Cloud Security Desk
Technical guideSource-based analysis

Validate the judge before trusting automated AI evaluations

Managed agent evaluators score your traces with a second model. Measure that judge against your own reviewers with chance-corrected agreement, probe its biases, and pin the judge model before its pass rate gates a release.

AI systems · Amazon Web Services / Microsoft Azure / Google Cloud · By Cloud Security Desk
Technical guideSource-based analysis

Keep semantic response caches inside authorization scope

A semantic cache returns whatever its nearest stored prompt earned. Put tenant, user or entitlement scope, template and model into an exact-match partition, close the write path, and treat thresholds as a correctness setting.

AI systems · Microsoft Azure / Google Cloud / Redis / GPTCache / Amazon Web Services · By Cloud Security Desk
Technical guideSource-based analysis

Give a SaaS vendor an AWS role without creating a confused deputy

A vendor's cross-account role is safe only if your trust policy demands an exact external ID, the vendor's backend generates and enforces it per tenant, and your reviews catch drift and offboard the role.

Identity & access · Amazon Web Services · By Cloud Security Desk

Questions answered

  1. What is in the Cloud Security Desk publication catalogue?

    The catalogue contains open-access research reports, visual briefs, technical guides, field notes, and research notes about cloud-control evidence.

  2. Can publications be filtered by topic or provider?

    Yes. The catalogue can be filtered by search terms, topic, format, cloud provider, author, and year.

    Supporting context

    Filtered and search views help readers navigate; the stable publication and topic pages remain the canonical research records.

  3. How is demonstration material identified?

    Catalogue cards and publication pages label illustrative analyses so synthetic scenarios and values are not mistaken for observed findings.

  4. Do publications include accessible evidence?

    Where applicable, publications include keyboard-accessible charts, text descriptions, source and method notes, numbered references, and downloadable datasets.

  5. Each publication is assigned to one or more control topics and links to related work, while the topic index groups research around a shared security boundary.

    Supporting context
  6. How often does the catalogue change?

    The catalogue updates when a new immutable publication edition goes live; the RSS feed carries the newest public releases.

    Supporting context