The Cyber Toolchain — issue 084, September 24, 2026

The Website Discovery list gains discovery date filtering, exact-match and exclusion text filters, and multi-select risk score filtering.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠ Clicking a count in the list now filters results to show only entries matching that value.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

Adds a Code Quality > Code Coverage section showing per-repository, per-file, and per-line test coverage alongside existing code quality findings.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠ Accepts LCOVLCOVAn open-source front end to GCC's gcov coverage tool that collects line and branch hit counts from instrumented C/C++ code and renders them as HTML reports, giving cyber tools a standard way to measure test coverage. and Cobertura report formats with native support for Jest and pytest-cov outputs, and provides a GitHub Action and a GitLab CI component to upload reports from CI pipelines with no extra token or self-hosted component required.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠ Supports merging split coverage reports from multiple services into a single view when multiple files are passed in one upload, integrates with GitHub, GitLab, Bitbucket, and Azure DevOps (plus a custom integration path for others), offers Code Tree and File List browsing modes, and tracks coverage over time with a per-repository trend graph updated on each upload.

Deep Review is an agentic PR reviewer that reasons across the full codebase — including middleware, global config, and linked repositories, not just the diff — flagging bugs and unintended side effects in addition to security findings. It saves accepted-risk and false-positive decisions to memory so future reviews reflect prior feedback, and runs after existing PR checks like SASTSASTStatic application security testing, analyzing source code or compiled code for vulnerabilities without running the application. It helps developers find problems before deployment. and dependency checks to build on rather than repeat their findings. It's enabled per repository with configurable linked repos and a review level setting, billed at 1 credit ($1) per PR with unlimited re-reviews on the same PR and a configurable monthly spend cap set via Aikido Wallet.

A Malware Firewall, embedded in Semgrep Guardian via Agent admin, intercepts dependency requests at the network level between the package manager and public registries and blocks malicious packages before they reach developer machines, requiring no per-device install.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠ It is currently in private beta.

The GitHub Copilot app adds per-project sandbox settings covering filesystem access (additional read/write, read-only, and denied folder lists), outbound internet and local network controls, and Git/GitHub CLI credential access for local repository and working tree sessions.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠ A new /sandbox on command enables sandboxing mid-session without changing the project default, and if the OS cannot enforce the requested policy the sandboxed shell errors out rather than silently running unsandboxed.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

$ /sandbox on

A new code review page under profile → Copilot settings (available on every plan, including Business and Enterprise) adds separate toggles for automatic reviews on PR creation, new pushes, and draft PRs you create or coauthor, plus a personal default review effort setting (Lite or Balanced) that applies to all requested and automatic reviews while still allowing per-request overrides from the Reviewers panel. Enterprise administrators can also set an enterprise-wide default review effort (Lite, Balanced, or GitHub default) for organization-owned repositories, with organization- and repository-level overrides still permitted.

⤢ ⤢

The new snyk studio install --experimental command configures Snyk Studio inside AI coding tools including Cursor, Claude Code, Codex, Copilot, Gemini, Kiro, and Windsurf.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠ Once installed, it scans code generated by these tools in the background as it is written.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

$ snyk studio install --experimental

snyk fix --agentic gains three new flags: --severity-filter to limit automated fixes to chosen severity levels, --breakability-filter to restrict Open Source upgrades to those matching a chosen breakability level, and --exclude-ids to skip specific issue IDs while fixing everything else.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

$ snyk fix --agentic --severity-filter=critical,high --exclude-ids=SNYK-JS-12345,SNYK-JS-67890

$ snyk fix --agentic --breakability-filter=low

snyk fix --agentic no longer automatically reverts failed fix attempts on upgrade; changes are now kept in place for review by default. Add --enable-revert to restore the previous automatic rollback behavior.

$ snyk fix --agentic --enable-revert

Supports per-agent LLM configuration across each of Signal AI's six agents, with provider, model, and endpoint independently selectable — supporting OpenAI, Azure OpenAI, AWS Bedrock, GCP Vertex AI, or any OpenAI-compatible proxy with automatic provider detection.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

Introduces central rule configuration in Polaris, letting security teams define rule profiles once and apply them at the organization, application, or project level via hierarchical inheritance, automatically enforced across CI pipelines, CLI scans, and integrations.

⤢

Extends on-premises SCM support to self-hosted GitLab, Azure DevOps Server, and Bitbucket Server/Data Center (in addition to previously released self-managed GitHub), enabling repository discovery, bulk onboarding, SASTSASTStatic application security testing, analyzing source code or compiled code for vulnerabilities without running the application. It helps developers find problems before deployment./SCA scan triggers, and pull request–driven remediation workflows.

⤢

Adds async SCA notifications — email alerts for new critical- or high-severity open source vulnerabilities, or reclassifications of existing vulnerabilities to critical or high, with project- and branch-level subscription controls.

⤢

Static Analysis now supports excluding Java .jar files and excluding specific Java classes from scans through the veracode.json configuration file.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

The Python scanner gains improved taint tracking, cross-file analysis, and pattern recognition, plus added support for analyzing PySpark, Flask, Django, and FastAPI applications.

Adds taint sinks for the send_from_directory and send_file functions in Python, adds taint propagation support for the Swift standard library, and expands .NET taint sink coverage for CWECWECommon Weakness Enumeration, a catalog of recurring software and hardware weakness types. It describes classes of mistakes, such as injection, rather than individual vulnerability instances.-80 to detect System.Web.UI.Page.Title and System.Web.UI.HtmlControls.HtmlHead.Title.

The Okta detector now detects and verifies OAuth 2OAuth 2An open authorization framework (RFC 6749) that lets a service grant limited access to a user's resources on another service without exposing credentials, giving cyber tools a standard way to authenticate against third-party APIs..0 client credentials (client_id + client_secret) in addition to the existing SSWS API token detection.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠ Verification uses the client_credentials grant against https://{domain}/oauth2/default/v1/token.

Adds per-detector verification wall-time tracking to the verification cache via an optional DetectorMetricsReporter interface, enabling metrics backends to record remote verification latency broken down by detector type.

Adds tls_key_exchange_preferences to server, agent, and proxy listener config to enforce pure post-quantum TLS key exchange algorithms (SecP256r1MLKEM768, SecP384r1MLKEM1024, X25519MLKEM768, MLKEM1024).⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

listener "tcp" {
  address = "0.0.0.0:8200"
  tls_cert_file = "/etc/openbao/tls/server.crt"
  tls_key_file  = "/etc/openbao/tls/server.key"
  tls_key_exchange_preferences = "X25519MLKEM768,MLKEM1024"
}

Adds consistency_fallback_behavior, consistency_missing_header_forward, and consistency_max_index_wait listener configuration parameters, and supports X-Vault-Index and X-Vault-Inconsistent request headers (values fail, forward-active-node, await-state) to enforce client/server read consistency.

$ curl -s -H "X-Vault-Token: $BAO_TOKEN" \
  -H "X-Vault-Index: $WRITE_INDEX" \
  -H "X-Vault-Inconsistent: await-state" \
  https://openbao.example.com:8200/v1/secret/data/myapp

Adds BAO_PROXY_DISABLE_IDLE_CONNECTIONS and BAO_PROXY_DISABLE_KEEP_ALIVES to control idle connection and keep-alive behaviour, BAO_ADDR to specify the upstream BAO server address, BAO_CACERT/BAO_CAPATH for CA certificate verification, BAO_CLIENT_CERT/BAO_CLIENT_KEY for mutual TLSmutual TLSMutual TLS, an encrypted connection where both client and server present certificates to prove their identity. It helps services verify who they are communicating with., BAO_SKIP_VERIFY to disable TLS verification, BAO_TLS_SERVER_NAME to override the TLS server name, and BAO_MAX_RETRIES to configure maximum retries in OpenBao Proxy.

$ export BAO_ADDR=https://bao.internal:8200
export BAO_CACERT=/etc/bao/ca.pem
export BAO_CLIENT_CERT=/etc/bao/client.crt
export BAO_CLIENT_KEY=/etc/bao/client.key
export BAO_MAX_RETRIES=3
bao proxy -config=/etc/bao/proxy.hcl

$ export BAO_PROXY_DISABLE_IDLE_CONNECTIONS=true
export BAO_PROXY_DISABLE_KEEP_ALIVES=true
bao proxy -config=/etc/bao/proxy.hcl

PKI and Transit engines can now use KMS plugins via /sys/external-keys APIs to perform cryptographic operations without storing key material, with built-in Transit-backed key support and a kms-pkcs11 plugin for PKCS#11PKCS#11The standard programming interface software uses to ask an HSM, smart card or token to sign or decrypt with a key it holds. Supporting it lets a tool keep its keys in hardware.. KMS plugin configuration stanzas are reloaded on SIGHUP, respawning plugin clients as needed to enable runtime plugin upgrades.

Adds plugin_version=latest to allow auth and secret mounts to automatically request the latest installed version of an external plugin, a bao plugin prune command to remove unused OCIOCIOpen Container Initiative standard, maintained by the Linux Foundation, defining the image format and runtime spec that container tools share so images built once run anywhere OCI-compliant. plugin downloads from disk, support for pinning OCI-based plugins by manifest digest (e.g. ghcr.io/openbao/openbao-plugin-secrets-aws@sha256:<digest>) instead of a sha256sum, and automatic determination of binary_name for OCI plugins from the image's ENTRYPOINT or CMD when unset.

Introduces ML-DSA (mldsa-44, mldsa-65, mldsa-87) key types in the Transit engine for post-quantum signing and verification, and adds ML-DSA (NIST FIPS 204FIPS 204NIST's main post-quantum digital signature standard (FIPS 204), formerly Dilithium. It is the planned replacement for RSA and ECDSA signatures on certificates and code.) support in the PKI engine for all CA, CSRCSRA Certificate Signing Request is a block of encoded data submitted to a certificate authority, containing a public key and identity details, so the CA can issue a signed TLS certificate without ever seeing the private key., and leaf certificate operations.

Adds PostgreSQL horizontal read scalability on the PostgreSQL storage backend, requiring ha_enabled = true and a reachable cluster_addr.

Adds a -field argument to bao status, bao namespace seal-status (which also gains -format), and bao token lookup.

Adds x_forwarded_for_client_cert_decoders Envoy decoder in auth/cert to process leaf certificates from the Envoy XFCC header.

Supports use_pss in issuers/generate/intermediate/:type PKI API calls for CSRs with PSS signatures from RSARSAThe oldest widely used public-key algorithm, for both encryption and signatures, with security resting on factoring large numbers. A large quantum computer would break it.-typed intermediates.

Adds tls_auto_reload configuration option to automatically reload TLS certificate and key files on change without requiring SIGHUP.

The pkcs11, alicloudkms, awskms, azurekeyvault, gcpckms, and ocikms seals are no longer built-in and must be installed as external plugins.

The built-in KerberosKerberosA network authentication protocol that uses tickets issued by a trusted service. Users and services can prove their identities without sending a password to each application. auth engine (auth/kerberos) and built-in LDAPLDAPA protocol for querying and modifying directory services, standardized under RFC 4511, commonly used to look up users, groups, and attributes in corporate identity stores like Active Directory. auth engine (auth/ldap) have been removed from the main OpenBao binary; both are now available in openbao-plugins.

The .../quit and .../cache-clear agent/proxy endpoints now require the X-Vault-Request header when require_request_header = true is set in listener configuration; calls omitting the header are rejected.

The flowcarp CLI adds --input, --output, and --model flags to build and apply statistical protocol models from PCAP files.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠ Custom models for proprietary or novel C2 protocols can be trained with flowcarp --output model:<label>:<tag>:<file>.fcp (e.g. flowcarp --input xyz.pcap --output model:MalwareXYZ:MALICIOUS:malwarexyz.fcp) and then applied to new captures with flowcarp --model malwarexyz.fcp --input new.pcap --output -. Classification distinguishes C2 traffic by packet sizes, timing, message-length distributions, request/response behavior, and protocol structure, covering 600+ protocols on the commercial license, and assigns a single protocol or malware-family label per flow rather than the multi-alert overlap common in traditional IDSIDSAn intrusion detection system, monitoring traffic or host activity for suspicious behavior and producing alerts. Detection alone does not necessarily block the activity.-based C2 classification.

$ flowcarp --input xyz.pcap --output model:MalwareXYZ:MALICIOUS:malwarexyz.fcp

$ flowcarp --model malwarexyz.fcp --input new.pcap --output -

$ flowcarp --input xyz.pcap --output model:MalwareXYZ:MALICIOUS:malwarexyz.fcp

$ flowcarp --model malwarexyz.fcp --input new.pcap --output -

A free demo web service at https://demo.flowcarp.com accepts raw PCAP uploads via curl --data-binary and returns a JSON response listing detected flows, protocols, and alerts, letting users classify traffic without running FlowCarp locally.

$ curl --data-binary @dump.pcap https://demo.flowcarp.com | jq

cube-node now runs on the host network by default in Kubernetes deployments, keeping sandbox network namespaces alive across Pod restarts (#1815).⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠ Switching an existing cluster to this mode requires draining compute nodes and passing the explicit hostNetworkChangeAck confirmation flag; without it the migration cannot proceed.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

Adds the cubeNode.network.mtu config key to pin guest MTU when the node's uplink NICNICA network interface card or adapter, the hardware that connects a computer to a network. It sends and receives packets over the link. MTU cannot be auto-detected, preventing silently broken sandbox networking on deployments with constrained overlay CNIs (#1674).

cubeNode:
  network:
    mtu: 1450

The Python SDK's sandbox.connect now accepts a timeout parameter to control connection deadlines, letting callers cap how long a connect attempt may block (#1352).

sandbox.connect(timeout=10)

Cubelet can now configure the template install path so large templates can be moved to the data disk (#1675). Component repository and template artifact storage is extended to support S3, local disk, PVC, and NFS backends (#1749).

Unity Catalog Scala UDFs declare required secrets in a SECRETS clause and run under the function owner's permissions, so callers need no direct secret access.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠ Requires environment_version set to 6 or above (including session-scoped UDFs on serverless compute), or on classic compute Databricks Runtime 19 or above with standard access mode; secret-accessing Scala UDFs are limited to serverless SQL warehouses among SQL warehouse types, with dedicated access mode compute unable to invoke them directly.

$ CREATE OR REPLACE FUNCTION my_catalog.my_schema.fetch_api_key()
RETURNS STRING
LANGUAGE SCALA
SECRETS ('my_scope' = 'my_key')
ENVIRONMENT_VERSION 6
AS $$
  dbutils.secrets.get(scope = "my_scope", key = "my_key")
$$;

The system.ai.genie_one_mcp MCP Service in Unity Gateway is now generally available, exposing Genie as a conversational tool over the Model Context ProtocolModel Context ProtocolModel Context Protocol, a standard for connecting AI applications to external tools and data sources. A compatible server exposes capabilities an assistant can use through a common interface. so any MCP client or agent (Claude, ChatGPT, Cursor) can query Databricks data with Unity Catalog permissions enforced on every request. The Beta endpoint https://<workspace-hostname>/api/2.0/mcp/genie is deprecated and will be sunset on October 31, 2026; workloads must migrate to system.ai.genie_one_mcp before that date.

Scalar and Batch Unity Catalog Python UDFs are now GA, adding named handlers for row-at-a-time scalar UDFs, removal of the five-UDF-call-per-query limit, support for installing custom dependencies, and access to Unity Catalog service credentials and secrets from within UDF functions. Setting environment_version to 6 or above in the UDF definition unlocks these scalar capabilities on serverless compute, and scalar TIMESTAMP inputs are now aligned with PySpark behavior.

Adds the ug CLI to connect coding agents — including Claude Code, Codex, and Pi — to Unity Gateway, handling authentication, configuration, MCP server registration, and spend visibility in the terminal.

Adds ai_predict_class SQL function to train a classification model and predict class labels for rows where the target column is NULL, and ai_predict_value SQL function to train a regression model and predict numeric values for rows where the target column is NULL.

Unity Gateway unified tracing now captures coding-agent activity — tools and skills used — alongside model requests, viewable in Govern > Traces or via Genie.

Google SecOps adds a metadata.base_labels.ingestion_kv_labels field with key gcp_organization_id to UDM events ingested via Google Cloud direct ingestion, enabling multi-organization log source identification, detection rule scoping, and Data RBAC without additional sink or feed configuration.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

// In your YARA-L 2.0 rule, filter UDM events by org ID:
events:
  $e.metadata.base_labels.ingestion_kv_labels["gcp_organization_id"] = "123456789012"

After March 18, 2027, the MANDIANT_ACTIVE_BREACH_IOC, MANDIANT_FUSION_IOC, and OPEN_SOURCE_INTEL_IOC feeds will be removed; users must migrate to the GTI_IOC feed before that date.

Admins can download the Bindplane Enterprise (Google Edition) license key directly from the console via Settings > SIEMSIEMSecurity information and event management, collecting logs from many systems and correlating events to find suspicious activity. Analysts use it to investigate alerts and incidents. Settings > Collection Agents to onboard a new collection agent without raising a support ticket.

Effective January 25, 2027, write permissions will be removed from the chronicle.readonly OAuth scope; any workflow performing write operations must be updated to use the chronicle OAuth scope.

Adds Accessibility Appearance, Safety & Handling, and Liquid Glass settings to Apple Business profiles, letting admins skip those setup assistant steps during device enrollment.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠ Note that Accessibility Appearance and Safety & Handling apply to iPhones and iPads only.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

The Helm chart adds tolerations.<component> keys plus a tolerations.defaults fallback in Helm values, letting operators configure tainted-node tolerations (control-plane, GPUGPUA graphics processing unit, a processor that performs many calculations in parallel. AI workloads use GPUs to speed up model training and inference., spot/preemptible) for individual components rather than the whole deployment.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

$ helm upgrade \
  --values overrides.yaml \
  --namespace trendmicro-system \
  trendmicro \
  oci://public.ecr.aws/trendmicro/container-security/trendmicro-container-security

Helm installation now uses oci://public.ecr.aws/trendmicro/container-security/trendmicro-container-security as the primary installation source, with the GitHub container registry retained as the alternative.

$ helm upgrade \
  --values overrides.yaml \
  --namespace trendmicro-system \
  trendmicro \
  oci://public.ecr.aws/trendmicro/container-security/trendmicro-container-security

Adds Wayfinder Threat Hunting for Cloud with continuous, expert-led hunting across AWS, Azure, and GCP control-plane activity, enabled via existing Singularity Marketplace plugins for each cloud provider.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠ Coverage includes cloud control-plane abuse, IAMIAMIdentity and Access Management — a framework of policies and services for controlling who can authenticate to systems and what resources they are authorized to use, letting cyber tools enforce least-privilege and audit access. privilege escalationprivilege escalationGaining permissions beyond those an account or process initially had. It can turn limited access into administrator control or access to another user's resources., unauthorized access, and data exfiltrationdata exfiltrationMoving data out of a system or organization without authorization. It can happen through network transfers, removable media or other channels. — specifically hunts for IAM user enumeration, S3 bucket reconnaissance, root account logins, AKSAKSAzure Kubernetes Service, Microsoft's managed Kubernetes offering, which lets cyber tools offload cluster provisioning and scaling to Azure so engineers focus on workload logic instead of control-plane operations. cluster-admin credential access, suspicious IAM policy changes, AMI deregistration, telemetry destruction, and cross-tenant delegation changes. Curated indicators and behavioral rules are mapped to MITRE ATT&CKMITRE ATT&CKA publicly maintained knowledge base of adversary tactics and techniques observed in real attacks, maintained by MITRE. Cyber tools use it as a common taxonomy to classify, map, and communicate threat behavior. techniques and findings are enriched with Purple AI summaries for triage, integrating threat intelligence from both SentinelOne and Google Threat Intelligence across endpoints, identities, and cloud in a single hunting workflow.

Four new Windows detections cover living-off-the-land and ClickFix techniques: Windows Finger.exe Connecting to a Remote Host identifies payload retrieval and remote connections through finger.exe, Windows For Loop Usage Within Cmd.exe To Execute Commands catches command execution via for /f loops (a common ClickFix technique), Windows Node.exe Executing JS Script In Immediate Folder flags JavaScript execution by node.exe from unusual directories, and Windows Process Accessing IronLanguages Repository On GitHub identifies processes reaching out to the IronLanguages GitHub repository.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

Three new detections target osascript misuse on macOS: Cisco NVM - Osascript Network Connection for a Long Duration identifies long-lived network connections initiated via osascript, MacOS Osascript Executing Interactive Shell catches osascript spawning interactive shells, and MacOS Osascript Executing JavaScript Code With ObjC flags JavaScript for Automation code using the Objective-C bridge via osascript.