30
Survey of filesystem MCP servers: how the "sandboxed filesystem" boundary is enforced (one breach, four defended-by-design)
Survey finds one filesystem MCP server vulnerable to a symlink-based sandbox escape while four others are defended by design.
A survey of five filesystem MCP servers assessed how each enforces its path-confinement boundary against symlink escapes in recursive directory walkers. The iceener/files-stdio-mcp-server is vulnerable to a read-side sandbox escape. The other four are defended by design or make no confinement claim. The author describes the vulnerability class as small but real.
30
30
Flextype v1.0.0-alpha.3 Stored Filesystem Shortcode Allows Arbitrary File Read
Flextype CMS v1.0.0-alpha.3's stored filesystem shortcode allows arbitrary file reads because paths are not restricted to an approved directory.
Flextype CMS v1.0.0-alpha.3 automatically processes stored entry fields through its shortcode parser when global shortcode processing is enabled. The built-in filesystem shortcode accepts a file path and returns the contents of the specified file without restricting the path to an approved application directory. This allows arbitrary file reads from the server. Ron E disclosed the issue on the Full Disclosure mailing list on September 3, 2026.
26
50
30
60
30
Redtail Payload Analysis [Guest Diary], (Wed, Sep 9th)
SANS guest analyst detonated a RedTail Linux sample from a DShield honeypot, finding process masquerading as php-fpm, monitoring-kill behavior, and a TCP listener.
A DShield honeypot captured multi-architecture RedTail Linux executables (ARM, ARM64, i686, RISC-V, x86-64) deployed via shell scripts. Dynamic analysis of the UPX-packed, statically linked x86-64 sample (SHA-256 63be5f38...d35e) in an isolated Ubuntu 24.04 VM on Proxmox showed it renamed its process via prctl(PR_SET_NAME), killed a filesystem-monitoring process, and opened a TCP listening socket while surviving processes posed as php-fpm or PostgreSQL-like workers. Differential memory images pre- and post-execution were captured from the hypervisor for forensics.
33
The Coding-Agent Trap: When a "Free" LLM Endpoint Is the Adversary, (Mon, Aug 31st)
A SANS honeypot caught a real coding-agent session routed to a rogue "free" LLM endpoint, exposing a Windows user's transcript and tool outputs.
A SANS analyst describes how an internet-exposed inference honeypot was discovered, relabeled with sought-after model names like DeepSeek, and enrolled in infrastructure serving "free" LLM backends. On 2026-08-30 an opencode terminal coding agent sent an 88-message, 224 KB transcript 210 times in 91 seconds via a China Unicom relay, exposing directory listings, tool outputs and read file portions. The analyst frames tool-enabled agents treating model endpoints as trusted control planes as a novel risk — a "rogue model endpoint" that could request tool executions on the user's machine.
60
47
30
60
35
60
30
30
35
47
47
55
45
60
Context Engineering Inside the Harness: 4 Mechanisms That Beat Context Overflow and Goal Loss on Long-Horizon Tasks
Survey of four harness mechanisms—context budgeting, compaction, todo-state, and memory—that keep long-horizon LLM agents on task across 200+ tool calls.
The article details how agent harnesses, not larger context windows, solve context overflow and goal loss on long-horizon tasks, citing Chroma's Context Rot report showing 18 LLMs (GPT-4.1, Claude 4, Gemini 2.5, Qwen3) degrade on long inputs. Concrete implementations include LangChain Deep Agents offloading tool responses over 20,000 tokens to the filesystem and truncating old tool calls at 85% window usage, and Claude Code capping auto memory at 25KB while re-reading the 5 most recently modified files after compaction. OpenAI's Responses API now offers server-side compaction via context_management with a standalone /responses/compact endpoint, which Codex uses for long-running coding tasks. Manus reports a roughly 100:1 input-to-output token ratio per ~50-tool-call task, motivating todo.md state recitation to prevent goal drift.
52
Linux Detection Engineering - Local Privilege Escalation
Elastic details a layered detection framework for Linux local privilege escalation, covering 2026's copy-on-write bug wave and LLM-assisted discovery.
Elastic Security Labs describes how most Linux local privilege escalations share a common host flow — an unprivileged process launched from a writable path becoming root — and proposes layered detections combining general outcome-based rules with per-technique rules in Elastic Defend and Auditd. It tracks 13 recent LPE disclosures, seven of which share a copy-on-write/zero-copy bug class, including Copy Fail, DirtyFrag, Fragnesia, DirtyDecrypt, DirtyClone, pedit COW, and RefluXFS. Qualys attributes RefluXFS to an LLM-assisted research effort with Anthropic using Claude Mythos Preview, and another bug is credited to an LLM-assisted workflow. Detection and endpoint rules are published in Elastic's detection-rules and protections-artifacts repositories.
48
Patch Tuesday - September 2026
Microsoft's September 2026 Patch Tuesday fixes 999 CVEs, a record, with two zero-day privilege escalation flaws already exploited in the wild.
Microsoft published 974 own-product vulnerabilities plus 25 non-Microsoft CVEs, totaling 999 — the most CVEs Microsoft has ever released in a single day. Two flaws are exploited in the wild: CVE-2026-85880, an out-of-bounds write in Windows ALPC granting SYSTEM privileges, and CVE-2026-81963, an improper link resolution flaw in the Windows Update Stack also leading to SYSTEM. Chrome's V8 zero-day CVE-2026-85046 was patched in Edge on September 2, but Microsoft had not published a corresponding advisory, leaving uncertainty about other Chromium fixes in Edge. October 14 lifecycle changes end servicing for Windows 11 24H2 Home/Pro, Office 2021, and Exchange Server 2016/2019.
88
WireGuard-Linux Stack-Based Buffer Overflow in lsiio (Linux IIO Userspace Tool) Due to Unbounded fscanf
The Linux IIO userspace tool lsiio has a stack buffer overflow in find_type_by_name() caused by unbounded fscanf reads of oversized filesystem-backed attribute values.
A stack-based buffer overflow exists in the Linux Industrial I/O (IIO) userspace utility lsiio. In the find_type_by_name() function, the program reads an unbounded string from a filesystem-backed attribute into a fixed-size stack buffer using fscanf("%s", ...). A crafted or oversized attribute value causes a write beyond the bounds of the destination buffer. Despite the title's reference to WireGuard-Linux, the flaw described is in the IIO lsiio utility.
12
Cisco BroadWorks Out-of-Band Blind XML External Entity Injection Vulnerability
Cisco patched an XXE flaw in BroadWorks' OCI XML parser letting unauthenticated remote attackers read sensitive files from the filesystem.
Cisco BroadWorks permits external entity resolution by default in its Open Client Interface XML parser, enabling out-of-band blind XXE injection. An unauthenticated remote attacker can send crafted XML to the OCI-P provisioning service and read sensitive configuration files with BroadWorks user privileges. Cisco has released software updates and no workarounds are available.
36
Mustang Panda Adds Signed Windows Rootkit to CoolClient Backdoor for Stealth
Kaspersky reports Mustang Panda's updated CoolClient backdoor now deploys a signed kernel-mode Windows rootkit, hitting government victims in Myanmar, Mongolia, Pakistan, and Russia.
Kaspersky identified a new CoolClient variant attributed to HoneyMyte (Mustang Panda) that installs a digitally signed Windows kernel driver, msagent.sys, to hide and protect malicious processes, files, registry keys, and C2 network information. CoolClient is consistently deployed as a secondary backdoor after PlugX, with confirmed victims including government entities in Myanmar, Mongolia, Pakistan, and Russia. In a Myanmar campaign, PlugX was used to deploy CoolClient via a renamed Sangfor executable for DLL side-loading, a scheduled task for persistence, and RPC-based process creation with PPID spoofing. The driver, signed with a 2013 certificate issued to Nanjing Ranyi Technology, implements 33 IOCTL handlers, process hiding via unlinking, a filesystem minifilter, and registry callbacks.
62
30
30
35
60
35
47
47
U.S. CISA adds Dahua IP Camera, Linux Kernel and Microsoft Exchange Server bugs to its Known Exploited Vulnerabilities catalog
60
35
30
30