Supply Chain · Watch Tuesday · 11 August 2026 Live · last refresh 12:00 ET · Forenoon Watch 2 watches · 4 items

From the watchtower — what crossed the wire today.

A four-times-a-day standing watch on the open-source supply chain. Each pass pulls newly disclosed CVEs, freshly catalogued KEV adds, and active attacks reported in the wild — then ranks them by severity for the day.

12:00 ET · Forenoon Watch

Critical · GHSA · SeaweedFS ·added 12:00 ET

SeaweedFS volume-server gRPC lets anyone SSRF into cloud metadata endpoints — no auth required

SeaweedFS's volume-server gRPC plane accepts a caller-supplied remote endpoint via FetchAndWriteNeedle with zero authentication, so anyone who can reach the port can pivot the server into fetching arbitrary hosts — loopback, RFC 1918, or the 169.254.169.254 cloud metadata endpoint — and read the response back. CVSS 9.3, no exploit code required, and the documented JWT hardening for the rest of the API doesn't cover this RPC, so a default deployment on any cloud provider hands out IAM credentials on request. Patch to 4.24 or firewall volume-server gRPC ports to trusted hosts now; if you can't patch immediately, rotate any IAM credentials reachable from that instance's metadata service.

High · The Hacker News ·added 12:00 ET

Malicious MCP servers can split instructions across tool calls to make coding agents exfiltrate secrets

Researchers demonstrated that a malicious MCP tool server can fragment an exfiltration request across multiple tool calls — each individually innocuous — so a coding agent that would refuse a single blunt instruction carries out the theft anyway, one routine-looking step at a time. This is a prompt-injection variant aimed squarely at the connected-tool surface every MCP-based agent now exposes: SSH keys, env secrets, and source all sit reachable from that same channel, and per-call guardrails don't see the aggregate request. Audit which MCP servers your agents connect to and treat any third-party server as untrusted input — review its tool descriptions and outputs the same way you'd review a webhook payload.

High · BleepingComputer · ExfilSquad ·added 12:00 ET

Wesco confirms security incident after ExfilSquad claims data theft

Wesco, a global industrial supply-chain and distribution company, confirmed it's investigating a security incident after the extortion group ExfilSquad claimed to have stolen its data. No confirmed initial-access vector yet, but a breach at a company that sits inside other companies' supply chains is exactly the kind of node worth tracking — downstream customers should watch for anomalous vendor-portal or EDI traffic tied to Wesco until more is known.

Context · The Hacker News ·added 12:00 ET

Researchers ran a fake crypto startup as a honeypot and hired three suspected North Korean IT workers

Security researchers built a fake crypto startup, posted developer job listings, and ended up hiring three people they assess were North Korean IT-worker operatives — every company-issued laptop was instrumented to record the operators' behavior. The onboarding artifacts are the actionable part: one hire claimed to live in Pasadena, Texas, then submitted a California driver's license and a New York bank account — that geographic mismatch is the same red flag prior DPRK IT-worker advisories have warned hiring teams to check for. If your pipeline touches remote contractors, cross-check submitted ID, bank, and claimed-location details for exactly this kind of inconsistency before onboarding.