This Week's Highlights — issue 009, August 28, 2026
THE CYBER TOOLCHAIN
NO. 009
The best tooling updates that shipped this week.
The week in numbers · Aug 21–27
Changes per week, last 12 weeks · Aug 21–27
Changes we read in that week's dailies, including docs and API changes. A gap means no daily was published.
Every feature release from the 1309 tools on our watchlist goes into the daily newsletter. Once a week we read the whole field side by side and choose our top picks, judged on two questions: how deep and complete is the single best capability in the release, and how much it changes what you can actually do.
A tool is judged on everything it shipped that week, so a project that releases daily gets credit for the sum — and still only takes one slot. We would rather run a short list than a padded one.
Focus: What features shipped and what problems they solve, plus launches, packaging and where the market is moving.
Leaves out: Command-line flags, API-only deltas, usage examples, and releases that are only refinements.
Focus: New security controls, new APIs, and product improvements you can put to work, plus anything an upgrade breaks.
Leaves out: Launch and marketing news, governance, pricing, and documentation changes with no operational effect.
Focus: Dashboards, reporting, and anything you can quantify: metrics, coverage, risk scores, compliance impact, vendor moves.
Leaves out: Settings, flags, API and command-line detail, library bumps, and minor releases with no reporting or posture angle.
Focus: API, SDK and MCP changes, new libraries, breaking changes in what you import, and cyber tooling.
Leaves out: Governance, SOC operations, and vendor launches with no developer surface.
This issue came out before editions were ranked, so it shows the Standard edition in issue order.
The picks in full
Trusted by 7M+ developers, SonarQube is the AI code review and code verification platform that catches quality and security issues before they reach production.
SonarQube's Hunter Agent is now GA on SonarQube Cloud, autonomously finding broken access control (IDOR, missing authorization, privilege escalation), business logic flaws, and auth/session weaknesses that SAST structurally cannot evaluate — validating each finding for exploitability before surfacing it as a normal SonarQube issue at 80–90% precision.
SonarQube launched Hunter Agent, a GA AI security agent on SonarQube Cloud's Enterprise plan that finds broken access control, business logic, and authentication flaws that traditional SASTSASTStatic application security testing, analyzing source code or compiled code for vulnerabilities without running the application. It helps developers find problems before deployment. cannot evaluate.
what's the number?
Each feature carries 0–100 for how completely the vendor documented it — not how big or important the work is. A major capability described in eight words scores low, and that is the finding.
- depth0–40 what it does and how it works or what changed
- specificity0–30 names real surfaces — APIs, flags, formats, limits, numbers
- actionability0–30 enough to go use it — a named endpoint, flag, or config key tops this; a UI path is a starting point
Code and config rank first by construction: a documented endpoint, flag or config key scores at the top of specificity and actionability, so it sorts above a feature described only in prose. Under 60 the notes go thin — everything below the “thinner coverage” line is thinner documentation, not smaller work. Hover any meter for that feature's three sub-scores.
Hunter Agent is a new AI security agent, GA on SonarQube Cloud (Enterprise plan), that detects broken access control (IDORIDORInsecure Direct Object Reference, a web vulnerability where an application exposes internal identifiers (IDs, filenames) without access checks, letting an attacker substitute another user's ID to reach their data., missing authorization checks, 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., CSRFCSRFCross-site request forgery, an attack that makes a signed-in browser send an unwanted request to a site. The site may accept the request using the victim's existing session.), business logic flaws (skipped workflow steps, missing rate limits), and authentication/session management flaws (session fixation, non-expiring sessions, weak password recovery, missing MFAMFAMulti-factor authentication, requiring proof from different categories, such as a password and a hardware key, to sign in. It reduces reliance on a password alone.) — vulnerability classes SAST cannot structurally evaluate. Every candidate finding is validated for exploitability by a dedicated subagent before surfacing, achieving 80–90% average precision, and each finding carries an auditable discovery path showing how the agent reached its conclusion. Confirmed findings appear directly as SonarQube issues in the existing triage workflow with no separate portal or install, can be triggered on-demand or scheduled weekly/monthly against a main branch, and are delivered asynchronously. It scales to codebases beyond 10 million lines via a structured multi-phase pipeline (Analyze, Explore, Validate, Synthesize) with incremental scanning that cuts large-repo re-scan time roughly in half, and works across the 40+ languages already supported by SonarQube with no additional configuration.
Aikido Security is a developer-focused security platform that scans code, dependencies, containers, and cloud infrastructure to detect and prioritize vulnerabilities for fast remediation.
Agentic Android (APK) pentesting installs a real build on a device and attacks it over ADB, with every finding exploit-confirmed and reproduction steps given as exact ADB commands — and it assesses the app plus its backend API in one run. The same week also shipped agentic Code Audit for logic flaws, code-based DSPM data-flow tracing, and real-time malware blocking on developer devices.
Aikido's biggest addition this window is agentic Android (APK) pentesting with ADBADBAndroid Debug Bridge, a command-line tool maintained by Google that lets a host machine communicate with an Android device over USB or TCP, giving cyber tools shell access, file transfer, and app control.-based exploit confirmation, alongside a wave of new AI agents: a Device Protection agent for company devices (now including Linux), LLMLLMA large language model, a neural network trained on large amounts of text to recognize patterns and generate language. Applications use it for tasks such as answering, summarizing and writing code.-powered Secrets AutoTriage and CVE Exploitability Analysis, code-based Data Exposure Audit (DSPMDSPMData security posture management, locating sensitive data and checking how it is protected and who can reach it. It helps teams prioritize risks around their data stores.), and pentest-grade Code Audit that needs no live environment. It also took GCP VM scanning to general availability and shipped smaller platform, reporting and API updates.
what's the number?
Each feature carries 0–100 for how completely the vendor documented it — not how big or important the work is. A major capability described in eight words scores low, and that is the finding.
- depth0–40 what it does and how it works or what changed
- specificity0–30 names real surfaces — APIs, flags, formats, limits, numbers
- actionability0–30 enough to go use it — a named endpoint, flag, or config key tops this; a UI path is a starting point
Code and config rank first by construction: a documented endpoint, flag or config key scores at the top of specificity and actionability, so it sorts above a feature described only in prose. Under 60 the notes go thin — everything below the “thinner coverage” line is thinner documentation, not smaller work. Hover any meter for that feature's three sub-scores.
Aikido agents install a real APK on a test device, log in as a user, and attack the app over ADB, producing exploit-confirmed findings with the exact ADB commands needed to reproduce each issue. Runs can simultaneously assess the Android app and its backend API in one report, cover unsafe WebView usage, deep-link and exported-component abuse, client-side authentication issues, injection flaws, broken access control, weak session handling, unsafe API behavior, IDORs, cross-tenant access, and authentication bypasses, and produce a recon report and attack plan with diagrams, threat scenarios, and the full agent reasoning trace (including attempts that weren't confirmed). Assessments are rightsized to app size rather than flat-fee, are currently white-box only (source repo access required alongside the APK, black-box support coming soon), and require certificate pinning, root detection, emulator detection, and RASP to be disabled in the uploaded build.
A new Device Protection agent provides real-time malware blocking across npm, PyPI, Maven, and NuGetNuGetMicrosoft's package manager for .NET, hosting a public registry of reusable libraries so tools can pull in dependencies without bundling or manually maintaining third-party code. package managers, powered by Aikido Intel, plus extension governance covering VS Code, OpenVSX, Chrome, and skills.sh extensions. A request-and-approval workflow pauses blocked installs and routes a one-click exception request to the admin Inbox inside Aikido, group-based policies let teams set different install and extension rules per team, and continuous device inventory gives an always-current view of every package and extension installed per device; the agent runs silently and only surfaces desktop notifications when a package is blocked or held for approval. Device Protection for Linux (beta) extends this to Linux desktops and servers, blocking malware at install time via .deb and .rpm packages.
Data Exposure Audit uses LLM agents to trace sensitive data from sources (request bodies, uploads, DB reads, config, app logic) into sinks (logs, third-party APIs, LLM prompts, caches, queues, storage), running entirely against already-connected repositories with no production data store access required. It detects credentials in URLs (tokens and API keys in query strings, redirects, and clone URLs), API over-exposure (endpoints returning more fields than the caller is entitled to, missing tenant ownership checks), unsafe logging and exports of unredacted PIIPIIPersonally Identifiable Information — data that can identify a specific individual, such as names, email addresses, or SSNs. Cyber tools flag or redact it to limit exposure and meet privacy regulations. and secrets, over-broad sharing of data with third parties and AI/LLM prompts and vector storesvector storesA database built to store embeddings and find the nearest ones to a query quickly. It is the retrieval half of most RAG systems., weak storage (unencrypted or only base64/MD5/SHA-1SHA-1A cryptographic hash function standardized by NIST that produces a 160-bit digest; tools use it for file integrity checks, though it is now considered cryptographically broken and unsuitable for collision-resistant applications. protected), incomplete deletion flows leaving downstream records, caches, and shadow copies intact, and over-permissioned 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. roles, OAuth scopes, and unauthenticated data stores. Each audit consumes credits from the Aikido Wallet.
Code Audit runs agentic, pentest-grade security reasoning directly on source code to find logic flaws (e.g. cross-tenant data leakage) that pattern-based scanners miss, with no staging URL, traffic replay, or agent deployment required. It surfaces code-based evidence, root cause analysis, and remediation guidance, and AutoFix can generate remediation PRs from findings in a few clicks. Pricing is shown upfront in credits before the audit starts, calculated from the size and complexity of the selected codebase, and it supports hard-to-test-by-URL targets including mobile apps, smart contracts, desktop apps, embedded software, and libraries.
Secrets AutoTriage is an LLM agent that runs after entropy detection to assess whether a discovered secret is a real, exposed credential by evaluating liveness, scope, and surrounding code context, then applies defined rules or LLM reasoning to automatically ignore or escalate findings. CVE Exploitability Analysis is an LLM agent that evaluates whether a CVE is practically reachable and exploitable given how the codebase uses the affected package, and can automatically upgrade, downgrade, snooze, or ignore the finding; this assessment now also supports containers. Every agent run is logged with a short rationale explaining what was analyzed and what action was taken, so teams can review or override decisions.
GCP VM Scanning is now generally available, using an agentless setup that requires no SSH keys or agents, completing Aikido's VM scanning coverage across AWS, Azure, and GCP.
Zen Firewall agent releases are now available for Dotnet, Java, PHP, Ruby, Golang, and Node.
Adds the ability to request ignore of an issue on a PR/MR with admin approval, and the ability to ignore license issues by path.
Adds an OSCAL FedRAMPFedRAMPA US government program that standardizes cloud security authorization, letting vendors get a single assessment accepted across federal agencies instead of repeating the process for each one. DASTDASTDynamic application security testing, probing a running application from the outside for vulnerabilities. It checks behavior exposed through the application's interfaces. report format under Reports.
Package Health adds support for NuGet and JetBrains packages.
New API to bulk update GitHub PR configuration across repositories.
New API to request a hardened container image.
New VSCode IDE extension release.
Zscaler is a cloud-native security platform that provides secure internet access, zero trust networking, and threat protection for users, devices, and applications.
Zscaler Zero Trust Browser can now discover, monitor, and block WebMCP AI-agent tool calls in-page, with inline policies matching on tool name, provider origin, and argument content — DLP-style inspection and enforcement of an agentic surface that was previously invisible.
Zscaler added WebMCP Security Controls to its Zero Trust Browser, giving admins visibility into and inline control over AI agent tool calls on WebMCP-enabled sites.
what's the number?
Each feature carries 0–100 for how completely the vendor documented it — not how big or important the work is. A major capability described in eight words scores low, and that is the finding.
- depth0–40 what it does and how it works or what changed
- specificity0–30 names real surfaces — APIs, flags, formats, limits, numbers
- actionability0–30 enough to go use it — a named endpoint, flag, or config key tops this; a UI path is a starting point
Code and config rank first by construction: a documented endpoint, flag or config key scores at the top of specificity and actionability, so it sorts above a feature described only in prose. Under 60 the notes go thin — everything below the “thinner coverage” line is thinner documentation, not smaller work. Hover any meter for that feature's three sub-scores.
Inline policies match on any of the seven webmcp.* attributes and apply one of three effects — Allow, Monitor, or Block — intercepting AI agent tool calls before the site's handler runs. Rules support matching on tool identity (webmcp.tool.name), argument content (webmcp.tool.arguments), provider origin (webmcp.tool.provider_origin), and cross-origin or untrusted-content signals (webmcp.tool.is_cross_origin, webmcp.tool.has_untrusted_content_hint) for DLPDLPData Loss Prevention: a category of controls that inspect and block sensitive data from leaving an organisation via email, web, or removable media, giving cyber tools a policy enforcement layer over data in motion and at rest.-style, argument-level inspection of agent payloads. Blocked calls return a structured failure to the agent ('The Zscaler Zero Trust Browser blocked this action for security reasons') with the exact rule reason (e.g., webmcp.tool.name is equal to UPDATE_CART) logged as a Low-severity Block detection.
update_cart on any covered site, with the block surfaced inline to the user.webmcp.tool.name is equal to update_cart, effect: Block. The Detections view will show a Low-severity Block entry with Reason: webmcp.tool.name is equal to UPDATE_CART on every blocked invocation.webmcp.tool.arguments contains book, effect: Block. Calls to update_cart with payloads such as {"cart":{"line_items":[{"handle":"book-cat-graphic-tee-apple-blossom-26f12103","quantity":1}]}} are blocked and the full argument payload is captured verbatim in the detection record.Every WebMCP tool invocation is captured with full context — the seven webmcp.* attributes plus user identity, device posture, extension version, browser, IP, geolocation, and URL — and surfaced in the Zscaler admin console as a searchable, filterable, exportable detection in the same event stream as existing platform telemetry.
webmcp.* attribute is logged as an Informational/Allow detection without disrupting users.The Zero Trust Browser Extension enumerates every tool a WebMCP-enabled site registers before any can run, capturing seven attributes per tool — webmcp.tool.name, webmcp.tool.description, webmcp.tool.arguments, webmcp.tool_count, webmcp.tool.provider_origin, webmcp.tool.is_cross_origin, and webmcp.tool.has_untrusted_content_hint — to build a structured inventory of the agentic surface per site and user.
JFrog is a DevOps platform providing software supply chain solutions including artifact management, CI/CD integration, and security scanning.
JFrog Agent Package Resolution routes AI coding-agent dependency downloads (npm, PyPI, Maven, Go, Docker, Helm, NuGet) through Artifactory Curation automatically — including transitive and postinstall pulls — closing the supply-chain blind spot that Claude Code and Cursor sessions reopened, with full audit trails.
JFrog opened a new Preview early-access track for upcoming capabilities, shipped Agent Package Resolution to route AI coding agents' dependency downloads through Artifactory governance, and added On-Behalf-Of token exchangetoken exchangeAn OAuth 2.0 protocol extension (RFC 8693) that lets a service trade one token for another with different scopes, audience, or type, enabling delegation and impersonation across service boundaries without re-authenticating the user. so Amazon Bedrock AgentCore Gateway can propagate per-user identity into Artifactory.
what's the number?
Each feature carries 0–100 for how completely the vendor documented it — not how big or important the work is. A major capability described in eight words scores low, and that is the finding.
- depth0–40 what it does and how it works or what changed
- specificity0–30 names real surfaces — APIs, flags, formats, limits, numbers
- actionability0–30 enough to go use it — a named endpoint, flag, or config key tops this; a UI path is a starting point
Code and config rank first by construction: a documented endpoint, flag or config key scores at the top of specificity and actionability, so it sorts above a feature described only in prose. Under 60 the notes go thin — everything below the “thinner coverage” line is thinner documentation, not smaller work. Hover any meter for that feature's three sub-scores.
Agent Package Resolution (preview), part of the JFrog Agent Plugin, routes AI coding agent dependency downloads (Claude Code, Cursor) through JFrog Artifactory instead of public registries, covering npm, PyPI, Maven, Go, Docker Hub, Helm, and NuGetNuGetMicrosoft's package manager for .NET, hosting a public registry of reusable libraries so tools can pull in dependencies without bundling or manually maintaining third-party code.. Session steering automatically supplies the Artifactory URL for each governed package type at the start of every agent session with no per-project configuration, while the new jf setup command persistently binds the local package manager to an Artifactory repository for a given project across sessions and tools. JFrog Curation policy enforcement is extended server-side to all agent-driven downloads — blocking malicious packages, unapproved licenses, and policy-violating packages regardless of whether session steering or jf setup is active — and covers indirect/transitive installs such as postinstall scripts, pip install -r requirements.txt, and npm ci. Every agent-driven download, block, and policy decision is captured in audit trails alongside existing human-driven build logs.
$ jf setup
Artifactory now supports On-Behalf-Of (OBO) OAuth 2.0OAuth 2.0An open authorization framework (IETF RFC 6749) that lets a service grant limited API access to a third party without sharing credentials, giving cyber tools a standard way to authenticate against external platforms. token exchange (RFC 8693RFC 8693An IETF standard defining the OAuth 2.0 Token Exchange flow, letting a service swap one token for another with different scopes, audience, or type — enabling delegation and impersonation without re-authenticating the user.) between Amazon Bedrock AgentCore Gateway and JFrog Artifactory Access, so AI agent requests reach Artifactory authenticated as the originating user rather than a shared service account, preserving audit fidelity, least privilege, and containment. AgentCore Gateway registers Artifactory's MCPMCPModel 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. server as a target configured with onBehalfOfTokenExchangeConfig, exchanging each inbound user JWTJWTAn open standard (RFC 7519) for encoding claims as a signed, optionally encrypted JSON object passed between parties. Cyber tools use it to verify identity and permissions without querying a central session store. for a short-lived per-user JFrog token; the listingMode field (requiring a recent AWS CLI) enables dynamic tool listing so discovery runs under the user's token instead of a shared credential. JFrog Access OIDCOIDCOpenID Connect, a sign-in protocol built on OAuth 2.0. An application uses identity information from a trusted provider instead of maintaining a separate password for each user. identity mappings integrate with AgentCore Identity by trusting the external IdP issuer ($IDP_ISSUER) and mapping verified access-token claims (e.g. preferred_username, team) to JFrog users, groups, or project roles via the JFrog OIDC integration API, requiring the token's aud claim to equal the AgentCore Gateway URL ($GATEWAY_URL) and iss to match $IDP_ISSUER. This approach is compared against header whitelisting and three-legged OAuth, with OBO being the only option delivering verifiable per-user identity without a second login; it requires JFrog EnterpriseX or higher.
JFrog launched a Preview early-access track offered as Closed Preview (invite-only) or Open Preview (all eligible customers), both free to opt into. Customers activate it in the MyJFrog portal under Settings > Preview by enabling the Enable Preview toggle, accepting the Preview Terms of Service in the Enable JFrog Preview dialog, and then selecting specific preview features to activate for their subscription.
DefectDojo Pro's Sensei adds CSPM: bulk-onboard AWS, Azure, and GCP accounts for cloud misconfiguration scanning with Fix in Cloud direct remediation through the provider API — gated by a human-approval dialog and a revert button — alongside a new EPSS/KEV-enriched Vulnerability Explorer for one-row-per-CVE prioritization.
DefectDojo Pro launched Sensei CSPMCSPMCloud security posture management, checking cloud resources for risky settings and policy violations. It helps teams find exposed services, overly broad permissions and other configuration gaps. for cloud misconfiguration scanning and one-click remediation across AWS, Azure and GCP, alongside a new Vulnerability Explorer with EPSSEPSSExploit Prediction Scoring System, an estimate of the probability that a published vulnerability will be exploited in the next 30 days. It helps prioritize investigation and patching./KEVKEVCISA's Known Exploited Vulnerabilities catalog, a list of vulnerabilities with evidence of exploitation in real attacks. It helps teams prioritize fixes for problems attackers already use. enrichment and a beta Root Cause Correlation clustering feature; the core platform also tightened endpoint permissions, added MFAMFAMulti-factor authentication, requiring proof from different categories, such as a password and a hardware key, to sign in. It reduces reliance on a password alone. disable verification, and expanded its documentation.
what's the number?
Each feature carries 0–100 for how completely the vendor documented it — not how big or important the work is. A major capability described in eight words scores low, and that is the finding.
- depth0–40 what it does and how it works or what changed
- specificity0–30 names real surfaces — APIs, flags, formats, limits, numbers
- actionability0–30 enough to go use it — a named endpoint, flag, or config key tops this; a UI path is a starting point
Code and config rank first by construction: a documented endpoint, flag or config key scores at the top of specificity and actionability, so it sorts above a feature described only in prose. Under 60 the notes go thin — everything below the “thinner coverage” line is thinner documentation, not smaller work. Hover any meter for that feature's three sub-scores.
DefectDojo Pro adds Cloud Connection onboarding to connect AWS Organizations, Azure subscriptions, and GCP projects in bulk using a shared read-only credential that auto-discovers and onboards all visible accounts at once. On-demand cloud posture scanning from the Sensei hub produces deduplicated, prioritized findings tied to a specific cloud resource (S3 bucket, security group, etc.) and updated in place on re-scan. A new Fix in Cloud action applies changes directly through the provider API using a separate write credential (independent of the read-only scan credential), gated by a mandatory human-approval dialog that shows the exact action, target resource, and required cloud permissions before execution; the existing IaCIaCInfrastructure as Code: the practice of defining and provisioning compute infrastructure through machine-readable config files rather than manual processes, giving cyber tools repeatable, version-controlled environment setup. pull-request remediation path is retained for resources managed as code, so Sensei opens a PR instead of touching the live resource when source exists.
DefectDojo Pro adds the Vulnerability Explorer, a deduplicated, one-row-per-vulnerability-ID view showing EPSS Score, EPSS Percentile, Known Exploited, Ransomware Used, and KEV Date columns plus a Total Findings count that links directly to the filtered findings list. A Sync KEV/EPSS data button pulls the latest CISA KEVCISA KEVA catalog maintained by the US Cybersecurity and Infrastructure Security Agency listing vulnerabilities confirmed to be actively exploited in the wild, giving cyber tools a vetted, authoritative prioritization signal. and FIRST EPSS data on demand in addition to an automatic daily sync. SLASLAA formal agreement between a service provider and customer that defines expected uptime, response times, and remedies for failures, giving cyber tools a contractual baseline for availability and support commitments. configuration now supports Cap by KEV due date, tightening a KEV-listed finding's deadline to CISA's official due date, and the view distinguishes unenriched vulnerabilities (blank column) from confirmed non-exploited ones (red ✗ under Known Exploited).
DefectDojo Pro adds a Beta Root Cause Correlation feature that groups findings sharing the same component-at-version, CVE, infrastructure resource, or weakness class at the same URL into ranked clusters showing active findings, affected assets, and aggregate risk. It is enabled by a superuser under Settings > Feature Flags.
DefectDojo's documentation was expanded to cover typed Custom Fields and their integration with Rules Engine 2.0, multi-Asset finding fan-out in Smart Upload workflows, AI Model Settings and supported LLMLLMA large language model, a neural network trained on large amounts of text to recognize patterns and generate language. Applications use it for tasks such as answering, summarizing and writing code. providers in Sensei, GMPGMPGNU Multiple Precision Arithmetic Library, a free library for arbitrary-precision integers, rationals, and floats; cyber tools use it when cryptographic operations require big-number math beyond native integer limits. over SSH support for OpenVAS/Greenbone connectors, customizable page layouts and Form Configuration in navigation, the Quick Export dialog for named reports, the rebuilt Asset hierarchy page, and the four-permission rule tier in Rules Engine 2.0. It also adds an RFC 9116 security.txt file to docs.defectdojo.com, adds Brazilian Portuguese, Simplified Chinese, and Italian translations of the core guides, and splits connector documentation into one page per connector.
Endpoint delete and rename operations are now scoped to the requesting product, restricting cross-product mutations that were previously possible.
Disabling MFA now requires entering a verification code, adding a confirmation step intended to protect account security.
Where each pick stands · Aug 21–27
Tools with changes in each pick's category over the last 12 weeks, including the pick itself.
Each pick's release rhythm · Aug 21–27
Changes per week for the last 12 weeks. The bright bar is this week; a dash marks a week with no daily.
Which categories moved · Aug 21–27
The thick bar is this week. The thin bar is the average of the four weeks before, counting only weeks with a daily.
Who shipped the most · Aug 21–27
Changes per tool this week. Volume is not a quality score.