https://cybertoolchain.io/tools/cycode New releases and features in Cycode, tracked by The Cyber Toolchain. en Fri, 23 Oct 2026 15:25:52 GMT https://cycode.com/blog/node2shell-cve-2026-111497/ https://cycode.com/blog/node2shell-cve-2026-111497/ Fri, 23 Oct 2026 15:25:52 GMT Cycode adds five CVE-2026-111497 queries to the Threat Intelligence dashboard, covering Node.js, Bun, .NET, and Python runtimes across Dockerfiles and built images⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠ • A new 'services that accept Brotli bodies' query in the Threat Intelligence dashboard finds Dockerfiles on a vulnerable Node.js or Bun image that also have, in the same directory, a dependency (Express 5, body-parser 2, Koa body parsers, or `@fastify/compress`) that decompresses Brotli request bodies.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠ • A new 'Dockerfiles on vulnerable runtimes' query finds Dockerfiles building on Node.js below 22.23.4, 24.22.1, or 26.12.1, end-of-life Node.js lines, or Bun below 1.4.4 — including digest-pinned images and `nvm install` or NodeSource setup lines that install a vulnerable version. • Two new container image queries in the Threat Intelligence dashboard inspect built images: one finds images whose Node.js or Bun runtime is below the fixed version, and another finds images shipping the distribution Brotli library in the affected range. • A new code dependencies query in the Threat Intelligence dashboard finds projects that depend on libraries bundling the vulnerable Brotli decoder — `brotli`, `brotlicffi`, and `curl_cffi` in Python, `com.aayushatharva.brotli4j` in Java, and the Ruby `brotli` gem — including indirect dependencies. • Each Threat Intelligence dashboard item for CVE-2026-111497 now links matched assets to their repository through the Context Intelligence Graph, so the owning team is clear. Full history on Cyber Toolchain: https://cybertoolchain.io/tools/cycode https://cycode.com/blog/emudhra-intermediate-ca-compromise/ https://cycode.com/blog/emudhra-intermediate-ca-compromise/ Fri, 23 Oct 2026 15:19:21 GMT Cycode adds three Threat Intelligence dashboard queries to find container images and Dockerfiles still trusting the compromised eMudhra intermediate CAs⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠ • Cycode's Threat Intelligence dashboard now includes three Context Intelligence Graph queries — images with unpatched `ca-certificates`, images with outdated runtimes (Node.js, Java, pip, certifi), and Dockerfiles on unpatched base images — each showing how many assets match and linking to the originating repository.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠ • Test whether a runtime still trusts the compromised CAs by connecting to `https://dv-g3b-emudhra.we-have-a-problem.com` and `https://dv-g3b-aaa-emudhra.we-have-a-problem.com` from inside containers and CI runners using `curl`, Python's `urllib.request`, or Node.js `fetch`; a certificate error means protected, a page load means still exposed. • Custom CA bundles pointed to via `SSL_CERT_FILE`, `REQUESTS_CA_BUNDLE`, `CURL_CA_BUNDLE`, `PIP_CERT`, or `-Djavax.net.ssl.trustStore` are never updated by the OS and must have the 16 compromised eMudhra intermediates blocked or removed manually. Full history on Cyber Toolchain: https://cybertoolchain.io/tools/cycode https://cycode.com/blog/emudhra-intermediate-ca-compromise/ https://cycode.com/blog/emudhra-intermediate-ca-compromise/ Fri, 23 Oct 2026 15:19:21 GMT Cycode adds three Threat Intelligence dashboard queries to surface eMudhra-compromised CA trust in container images, runtimes, and Dockerfiles.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠ • Adds 'Images with unpatched ca-certificates' query to the Threat Intelligence dashboard, showing container images whose `ca-certificates` package is older than the fixed version for Alpine/Chainguard (apk), Debian/Ubuntu (deb), and RHEL/Fedora (rpm), with results displaying image tags and installed version.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠ • Adds 'Images with outdated runtimes' query to the Threat Intelligence dashboard, identifying images carrying Node.js, Java, pip, or certifi below the fixed versions that bundle their own trust data independently of the base image. • Adds 'Dockerfiles on unpatched base images' query to the Threat Intelligence dashboard, flagging Dockerfiles that build on Node.js, Python, Go, Java, Bun, or Deno images below the fixed version, on an end-of-life line, or pinned by digest. • Surfaces eMudhra CA exposure across dependencies, container images, Dockerfiles/IaC, leaf certificates, CI/CD workflows, and developer machines via Context Intelligence Graph queries, with each result linking back to the originating repository. Full history on Cyber Toolchain: https://cybertoolchain.io/tools/cycode https://cycode.com/blog/custom-maestro-agents-agentic-workflows/ https://cycode.com/blog/custom-maestro-agents-agentic-workflows/ Thu, 22 Oct 2026 18:17:55 GMT Cycode adds Custom Maestro Agents — user-defined AI agents with configurable outcomes that plug into Agentic Workflows for automated violation triage.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠ • Adds Custom Maestro Agents: define an agent's analysis instructions, evidence to weigh, and up to three named outcomes (plus a reserved Failure outcome) in the AI agents library — covering use cases like upgrade impact analysis, credential validity, and code ownership mapping.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠ • Adds custom Maestro agents as steps in Agentic Workflows, where each outcome branch drives a downstream action — open a Jira ticket, message a Slack channel, set severity or status, add a label, open a fix pull request, or hand off to another agent. • Adds confidence thresholds per outcome branch (default 70%) so a branch executes only when the agent's confidence meets the configured minimum. • Adds 'Require human approval' toggle on any outcome branch's first action, surfacing the agent's recommendation in Required Approvals before anything executes. • Locks agent outcomes while a workflow uses an agent ('Stable logic'), preventing branch-breaking changes as the agent's instructions evolve. • Adds per-workflow token attribution on the AI usage dashboard, with a ready-made template that alerts when usage reaches 80% of quota. • Enables testing a custom agent against a real violation before publishing — returning the verdict, reasoning, and token consumption. Full history on Cyber Toolchain: https://cybertoolchain.io/tools/cycode https://cycode.com/blog/litellm-1-107-0-compromise/ https://cycode.com/blog/litellm-1-107-0-compromise/ Wed, 21 Oct 2026 20:35:46 GMT Cycode adds real-time Workstation Protection with cooldown policies and threat-intel blocking to stop malicious packages at install time.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠ • Introduces Cycode Workstation Protection (early access), deployable via MDM, which intercepts package installs on developer and CI machines in real time and applies a configurable cooldown policy to hold back recently published versions.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠ • Adds a threat intelligence feed check to Workstation Protection that blocks known-malicious package versions independently of the cooldown window. • Extends SBOM and AIBOM generation to cover AI/LLM dependencies (AIBOMs), surfacing transitive pulls like LiteLLM from agent frameworks such as DSPy and CrewAI. • Extends container scanning to detect poisoned proxy images on floating tags (such as `latest`, `main-latest`, `main-stable`) in connected registries. • The Context Intelligence Graph links a finding to its repo, image, pipeline, and team, enabling a single query to list all locations where a specific package version appears and identify owners. • Cimon, Cycode's eBPF-based build monitor, watches network calls, file writes, and processes on CI runners to detect supply-chain payload behaviors such as `.pth` file writes into `site-packages`. Full history on Cyber Toolchain: https://cybertoolchain.io/tools/cycode https://cycode.com/blog/cycode-hugging-face/ https://cycode.com/blog/cycode-hugging-face/ Tue, 29 Sep 2026 14:26:22 GMT Cycode integrates with Hugging Face to inventory, enrich, and enforce governance over every AI model your org publishes or uses.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠ • Adds a Hugging Face integration that connects to your Hugging Face organizations via a read-only token and syncs only the organizations you select, with an initial full pull followed by daily refreshes.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠ • Surfaces a unified AI model inventory combining models detected in your code repositories with models published by your connected Hugging Face organizations, including private org models that never appear in a repo. • Enriches each model with governance metadata pulled from Hugging Face: license, public/private visibility, gated-download status, base-model lineage, parameter count and precision, file types and serialization formats, downloads, likes, Spaces usage count, and benchmark availability. • Introduces a per-model authorization workflow — mark each model as `Not reviewed`, `Authorized`, or `Unauthorized` directly from the inventory — with `Unauthorized` models automatically raising violations and every status change written to the audit log. • Enables namespace filtering in the inventory so teams can review all models published by a single provider or Hugging Face organization at once. Full history on Cyber Toolchain: https://cybertoolchain.io/tools/cycode https://cycode.com/blog/introducing-workstation-protection/ https://cycode.com/blog/introducing-workstation-protection/ Wed, 23 Sep 2026 11:33:09 GMT Cycode Workstation Protection adds real-time package install enforcement with cooldown policies and malicious package threat intelligence.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠ • Adds Workstation Protection, a new capability that intercepts package install commands in real time to enforce cooldown (release-age gating) policies and block known malicious packages before they install on developer workstations.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠ • Deploys Workstation Protection via Mobile Device Management (MDM) with no new console, no new infrastructure, and no change to developer install workflows. • Checks every package install against a continuously updated threat intelligence feed to catch malicious packages that have aged past cooldown windows, including typosquatting, slopsquatting, and persistent compromised maintainer releases. • Enforces cooldown policies at the device level, so gating applies regardless of which package manager version a developer or coding agent is running or whether a local config file was configured correctly. • Extends ADLC Security coverage upstream to the developer workstation, unifying workstation package enforcement with existing AI Bill of Materials, prompt guardrails, pipeline integrity, and secret scanning into the Context Intelligence Graph. Full history on Cyber Toolchain: https://cybertoolchain.io/tools/cycode https://cycode.com/blog/introducing-workstation-protection/ https://cycode.com/blog/introducing-workstation-protection/ Wed, 23 Sep 2026 11:33:09 GMT Cycode Workstation Protection adds real-time package install enforcement with cooldown policies and malicious package threat intelligence.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠ • Adds Workstation Protection, a new capability that intercepts package install commands in real time to enforce cooldown (release-age gating) policies and block known malicious packages before they install on developer workstations.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠ • Deploys Workstation Protection via Mobile Device Management (MDM) with no new console, no new infrastructure, and no change to developer install workflows. • Checks every package install against a continuously updated threat intelligence feed to catch malicious packages that have aged past cooldown windows, including typosquatting, slopsquatting, and persistent compromised maintainer releases. • Enforces cooldown policies at the device level, so gating applies regardless of which package manager version a developer or coding agent is running or whether a local config file was configured correctly. • Extends ADLC Security coverage upstream to the developer workstation, unifying workstation package enforcement with existing AI Bill of Materials, prompt guardrails, pipeline integrity, and secret scanning into the Context Intelligence Graph. Full history on Cyber Toolchain: https://cybertoolchain.io/tools/cycode https://cycode.com/blog/introducing-workstation-protection/ https://cycode.com/blog/introducing-workstation-protection/ Wed, 23 Sep 2026 11:33:09 GMT Cycode Workstation Protection adds real-time package install enforcement with cooldown policies and malicious package threat intelligence.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠ • Adds Workstation Protection, a new capability that intercepts package install commands in real time to enforce cooldown (release-age gating) policies and block known malicious packages before they install on developer workstations.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠ • Deploys Workstation Protection via Mobile Device Management (MDM) with no new console, no new infrastructure, and no change to developer install workflows. • Checks every package install against a continuously updated threat intelligence feed to catch malicious packages that have aged past cooldown windows, including typosquatting, slopsquatting, and persistent compromised maintainer releases. • Enforces cooldown policies at the device level, so gating applies regardless of which package manager version a developer or coding agent is running or whether a local config file was configured correctly. • Extends ADLC Security coverage upstream to the developer workstation, unifying workstation package enforcement with existing AI Bill of Materials, prompt guardrails, pipeline integrity, and secret scanning into the Context Intelligence Graph. Full history on Cyber Toolchain: https://cybertoolchain.io/tools/cycode https://cycode.com/blog/introducing-workstation-protection/ https://cycode.com/blog/introducing-workstation-protection/ Wed, 23 Sep 2026 11:33:09 GMT Cycode Workstation Protection adds real-time package install enforcement with cooldown policies and malicious package threat intelligence.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠ • Adds Workstation Protection, a new capability that intercepts package install commands in real time to enforce cooldown (release-age gating) policies and block known malicious packages before they install on developer workstations.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠ • Deploys Workstation Protection via Mobile Device Management (MDM) with no new console, no new infrastructure, and no change to developer install workflows. • Checks every package install against a continuously updated threat intelligence feed to catch malicious packages that have aged past cooldown windows, including typosquatting, slopsquatting, and persistent compromised maintainer releases. • Enforces cooldown policies at the device level, so gating applies regardless of which package manager version a developer or coding agent is running or whether a local config file was configured correctly. • Extends ADLC Security coverage upstream to the developer workstation, unifying workstation package enforcement with existing AI Bill of Materials, prompt guardrails, pipeline integrity, and secret scanning into the Context Intelligence Graph. Full history on Cyber Toolchain: https://cybertoolchain.io/tools/cycode https://cycode.com/blog/agentic-code-scanning/ https://cycode.com/blog/agentic-code-scanning/ Tue, 01 Sep 2026 10:29:26 GMT Cycode adds Agentic Code Scanning as a fourth scanning dimension, catching authorization vulnerabilities that rule engines structurally cannot detect. • Adds Agentic Code Scanning (Dimension 4) — an agent-orchestrated layer that plans from a threat model, ranks code by vulnerability class, and disproves findings before surfacing them, enabling detection of logic-level flaws like broken access control that have no matchable pattern. • Introduces per-scan configuration of mode, model, and scope as three independent dials, so teams can target expensive reasoning only at code classes that rules cannot describe while keeping deterministic scanning on every commit. • Adds attack-chain synthesis that links related findings across files into a single multi-step exploit path, carrying data flow from untrusted source to dangerous sink and tagging the resulting risk category (e.g. Code Injection — Critical). • Adds Agentic Workflows that consume exploitability-qualified findings and automate remediation: teams define triggering events, agent action sequences, and confidence thresholds once, and the workflow fires automatically — routing findings to the owning engineer and returning fixes as reviewable pull requests with a full audit trail. • Implements a self-improving flywheel where Agentic Code Scanning findings that generalize are automatically converted into new deterministic SAST rules (Dimension 1/2), so a class caught once by the model re-runs at near-zero cost indefinitely. • Detected authorization CVEs (gitea `CVE-2025-68941` and next.js `CVE-2025-29927`) that Semgrep OSS, CodeQL, and Cycode's own rule engine all missed, across a ten-repository, six-language benchmark on an affordable open-weights reasoning model. • Integrates with Cycode's Context Intelligence Graph to ground attack chains in real call relationships, ownership, and reachability rather than speculative model inference. Full history on Cyber Toolchain: https://cybertoolchain.io/tools/cycode https://cycode.com/blog/agentic-code-scanning/ https://cycode.com/blog/agentic-code-scanning/ Tue, 01 Sep 2026 10:29:26 GMT Cycode adds Agentic Code Scanning as a fourth scanning dimension, catching authorization vulnerabilities that rule engines structurally cannot detect. • Adds Agentic Code Scanning (Dimension 4) — an agent-orchestrated layer that plans from a threat model, ranks code by vulnerability class, and disproves findings before surfacing them, enabling detection of logic-level flaws like broken access control that have no matchable pattern. • Introduces per-scan configuration of mode, model, and scope as three independent dials, so teams can target expensive reasoning only at code classes that rules cannot describe while keeping deterministic scanning on every commit. • Adds attack-chain synthesis that links related findings across files into a single multi-step exploit path, carrying data flow from untrusted source to dangerous sink and tagging the resulting risk category (e.g. Code Injection — Critical). • Adds Agentic Workflows that consume exploitability-qualified findings and automate remediation: teams define triggering events, agent action sequences, and confidence thresholds once, and the workflow fires automatically — routing findings to the owning engineer and returning fixes as reviewable pull requests with a full audit trail. • Implements a self-improving flywheel where Agentic Code Scanning findings that generalize are automatically converted into new deterministic SAST rules (Dimension 1/2), so a class caught once by the model re-runs at near-zero cost indefinitely. • Detected authorization CVEs (gitea `CVE-2025-68941` and next.js `CVE-2025-29927`) that Semgrep OSS, CodeQL, and Cycode's own rule engine all missed, across a ten-repository, six-language benchmark on an affordable open-weights reasoning model. • Integrates with Cycode's Context Intelligence Graph to ground attack chains in real call relationships, ownership, and reachability rather than speculative model inference. Full history on Cyber Toolchain: https://cybertoolchain.io/tools/cycode https://cycode.com/blog/agentic-code-scanning/ https://cycode.com/blog/agentic-code-scanning/ Tue, 01 Sep 2026 10:29:26 GMT Cycode adds Agentic Code Scanning as a fourth scanning dimension, catching authorization vulnerabilities that rule engines structurally cannot detect. • Adds Agentic Code Scanning (Dimension 4) — an agent-orchestrated layer that plans from a threat model, ranks code by vulnerability class, and disproves findings before surfacing them, enabling detection of logic-level flaws like broken access control that have no matchable pattern. • Introduces per-scan configuration of mode, model, and scope as three independent dials, so teams can target expensive reasoning only at code classes that rules cannot describe while keeping deterministic scanning on every commit. • Adds attack-chain synthesis that links related findings across files into a single multi-step exploit path, carrying data flow from untrusted source to dangerous sink and tagging the resulting risk category (e.g. Code Injection — Critical). • Adds Agentic Workflows that consume exploitability-qualified findings and automate remediation: teams define triggering events, agent action sequences, and confidence thresholds once, and the workflow fires automatically — routing findings to the owning engineer and returning fixes as reviewable pull requests with a full audit trail. • Implements a self-improving flywheel where Agentic Code Scanning findings that generalize are automatically converted into new deterministic SAST rules (Dimension 1/2), so a class caught once by the model re-runs at near-zero cost indefinitely. • Detected authorization CVEs (gitea `CVE-2025-68941` and next.js `CVE-2025-29927`) that Semgrep OSS, CodeQL, and Cycode's own rule engine all missed, across a ten-repository, six-language benchmark on an affordable open-weights reasoning model. • Integrates with Cycode's Context Intelligence Graph to ground attack chains in real call relationships, ownership, and reachability rather than speculative model inference. Full history on Cyber Toolchain: https://cybertoolchain.io/tools/cycode https://cycode.com/blog/introducing-agentic-code-scanning/ https://cycode.com/blog/introducing-agentic-code-scanning/ Tue, 01 Sep 2026 10:29:26 GMT Cycode adds Agentic Code Scanning: a four-dimension system that orchestrates rules, AI triage, and agentic reasoning to catch authorization flaws no rule engine can express. • Adds Agentic Code Scanning (Dimension 4) — an orchestration layer that plans from a threat model, ranks what the model reads per vulnerability class, and disproves every finding before surfacing it, running above the three deterministic/AI tiers. • Introduces per-scan control over mode, model, and scope as three independent dials, with every scan recording what ran, when, on which code, and with which model for auditability. • Agentic reasoning promotes confirmed vulnerability mechanisms into new deterministic SAST rules (Dimension 1/2), so a class found by the expensive layer is caught at near-zero cost on every subsequent run, compounding coverage as repo count grows. Full history on Cyber Toolchain: https://cybertoolchain.io/tools/cycode https://cycode.com/blog/agentic-code-scanning/ https://cycode.com/blog/agentic-code-scanning/ Tue, 01 Sep 2026 10:29:26 GMT Cycode adds Agentic Code Scanning as a fourth scanning dimension, catching authorization vulnerabilities that rule engines structurally cannot detect. • Adds Agentic Code Scanning (Dimension 4) — an agent-orchestrated layer that plans from a threat model, ranks code by vulnerability class, and disproves findings before surfacing them, enabling detection of logic-level flaws like broken access control that have no matchable pattern. • Introduces per-scan configuration of mode, model, and scope as three independent dials, so teams can target expensive reasoning only at code classes that rules cannot describe while keeping deterministic scanning on every commit. • Adds attack-chain synthesis that links related findings across files into a single multi-step exploit path, carrying data flow from untrusted source to dangerous sink and tagging the resulting risk category (e.g. Code Injection — Critical). • Adds Agentic Workflows that consume exploitability-qualified findings and automate remediation: teams define triggering events, agent action sequences, and confidence thresholds once, and the workflow fires automatically — routing findings to the owning engineer and returning fixes as reviewable pull requests with a full audit trail. • Implements a self-improving flywheel where Agentic Code Scanning findings that generalize are automatically converted into new deterministic SAST rules (Dimension 1/2), so a class caught once by the model re-runs at near-zero cost indefinitely. • Detected authorization CVEs (gitea `CVE-2025-68941` and next.js `CVE-2025-29927`) that Semgrep OSS, CodeQL, and Cycode's own rule engine all missed, across a ten-repository, six-language benchmark on an affordable open-weights reasoning model. • Integrates with Cycode's Context Intelligence Graph to ground attack chains in real call relationships, ownership, and reachability rather than speculative model inference. Full history on Cyber Toolchain: https://cybertoolchain.io/tools/cycode