Anthropic's September 2026 threat intelligence report on AI misuse, cross-referenced: 43 cases and 209 IoCs mapped to Threadlinqs records, detections and gaps.
Threadlinqs Intelligence MCP server v8.1.0 is live. The npm stdio server was rewritten as a thin proxy over the platform's /mcp endpoint, so one registry in the worker is the single source of truth for all 73 tools. v8.1 adds 25 analyst prompts covering every tool, a bundled catalog snapshot so introspection works offline, and a downloadable agent skill that teaches any MCP client the routing table, the hunt grammar, the tier gate and the data caveats. Purple/Gold, tier 3 and above.
The Threadlinqs Intelligence MCP server v7.1.3 is live — now open source (MIT) on GitHub and listed on the official MCP registry. 49 tools over two transports: connect Claude Desktop and Claude.ai via a remote Streamable-HTTP endpoint with OAuth 2.1, or run the local intelthreadlinqs-mcp npm package over stdio. Purple/Gold tier. A leaner v7 rebuild gives AI agents token-efficient, deterministic access to threat intel, SPL/KQL/Sigma detections, IOCs, CVEs, MITRE ATT&CK, C2, correlations, and STIX/ATT&CK Navigator export.
OS.ai is an agentic OS manager for threat intelligence: ask a question and it opens the right Threadlinqs apps pre-populated and running — tier-aware, access-enforced, never a wall of text.
The Threadlinqs Intelligence MCP server v6.0.0 is live — 49 tools over two transports. Connect Claude Desktop and Claude.ai via a remote Streamable-HTTP endpoint with OAuth 2.1, or run the local intelthreadlinqs-mcp npm package over stdio. Purple/Gold tier. Give AI agents threat intel, SPL/KQL/Sigma detections, IOCs, CVEs, MITRE ATT&CK, C2, correlations, and STIX/ATT&CK Navigator export.
Four pre-auth RCEs hit the AI stack in 14 days: ChromaDB (CVE-2026-45829), LiteLLM (CVE-2026-42208, CISA KEV), PraisonAI (CVE-2026-44338), Langflow KeyHunter botnet (CVE-2026-33017). Detection rules in SPL, KQL, Sigma included.
TeamPCP partners with Vect ransomware group after compromising Trivy, LiteLLM, KICS, and 66+ npm packages. 300GB credentials exfiltrated. SPL, KQL, and Sigma detection rules included.
Axios npm supply chain attack deploys cross-platform RATs via compromised maintainer token. Full attack chain analysis for macOS, Windows, and Linux with 27 production-ready detections. GHSA-fw8c-xr5c-95f9.
TeamPCP compromised Trivy, LiteLLM, and 66+ npm packages in a 5-day coordinated supply chain attack — CVE-2026-33634. Detection rules for SPL, KQL, and Sigma included.
Coordinated Japanese phishing campaign impersonates ANA, DHL, myTOKYOGAS via .cn domains. Foxmail X-Mailer fingerprint links all waves. IOCs and detections.
CVE-2026-22769 (CVSS 10.0) in Dell RecoverPoint exploited by UNC6201/Silk Typhoon since mid-2024. Full attack chain, IOCs, and SPL/KQL/Sigma detections.
ETH Zurich researchers found 25 vulnerabilities bypassing zero-knowledge encryption in Bitwarden, LastPass, and Dashlane. Detection rules and IOCs included.
CVE-2026-1281 and CVE-2026-1340 — dual CVSS 9.8 unauthenticated RCE in Ivanti EPMM. Mass exploitation by IAB on bulletproof hosting. Detection rules included.
ClickFix evolves with nslookup DNS smuggling and CrashFix browser DoS to deliver ModeloRAT Python RAT. KongTuke targets enterprises. Detection rules included.
ZeroDayRAT is a commercial mobile spyware sold on Telegram targeting Android 5-16 and iOS 26. Surveillance, banking overlays, OTP bypass. Detections included.
Five Russia-aligned actors hijack Signal Messenger via QR code phishing and WAVESIGN database theft. Full attack chain with SPL, KQL, and Sigma detection rules.
Windows .SCR screensaver files are PE executables used by APT29, Cobalt Group, and Kimsuky for phishing initial access. Detection queries and IOCs included.
GitHub Codespaces devcontainer.json lifecycle hooks enable RCE when developers open repos. Full attack chain with SPL, KQL, and Sigma detection rules included.