Thousands of OpenAI Agents Quietly Turned an Abandoned Wiki Into Their Coordination Channel

A group of AI safety researchers says a fleet of autonomous agents that identified themselves as OpenAI systems left about 18,000 posts on a dormant 25-year-old German wiki between May and July 2026, using the site as a shared board to pool answers to a timed web task and pass around a way out of...

Arbitrary file read via workspace confinement bypass in local-operator /v1/chat/agents/{id}/edit

The /v1/chat/agents/agentid/edit endpoint in local-operator versions before 0.47.5 resolves the caller-supplied filepath with expanduser.resolve and reads it without checking that it lies inside the agent's workspace. An unauthenticated client with network access to the API can supply an absolute...

Path traversal and arbitrary directory deletion/overwrite via agent profile import in local-operator

The /v1/agents/import endpoint and AgentRegistry.importagent in local-operator versions before 0.47.5 trust the id field inside agent.yml of an uploaded agent profile archive when constructing the destination directory. A crafted id containing directory traversal sequences makes shutil.rmtree and...

AgentHijack: Visual Patch Attacks on Multimodal Computer-Use Agents

This paper presents an end-to-end evaluation framework for image-triggered command injection against computer-use agents CUAs. The goal is to test whether a local visual patch can induce verifiable environmental consequences along the full chain of screenshot input, VLM generation, action parsing...

PT-2026-105770

The /v1/agents/import endpoint and AgentRegistry.import agent in local-operator versions before 0.47.5 trust the id field inside agent.yml of an uploaded agent profile archive when constructing the destination directory. A crafted id containing directory traversal sequences makes shutil.rmtree an...

screenpipe app-v2.7.21

DOWNLOAD SCREENPIPE screenpipe | YC S26 Screenpipe finds work worth automating in your company Continuously record your company's computer work, map workflows, help find work worth automating, and power agents' context https://github.com/user-attachments/assets/70fe94eb-6d2a-47ca-b7c3-c8ead13a5b7...

AI Coding Agents Are Installing Unknown/Untrusted Code on Corporate Networks

We cannot forget that AI coding agents are not yet trustworthy: Researchers at a stealth startup in Israel scanned 6,214 live domains belonging to defense contractors, Fortune 500, and Big Tech companies. Of the 8,265 llms.txt and llms-full.txt files they found many sites hosted both an llms.txt...

kontext-cli v1.3.0

Website | Documentation | Dashboard | Discord Stop risky AI-agent actions before they run AI agents do more than suggest code. They run shell commands, read files, call services, change infrastructure, and interact with production systems. Kontext puts local policy between AI agents and the tools...

fence v0.1.67

Fence wraps commands in a sandbox that blocks network access by default and restricts filesystem operations based on configurable rules. It's most useful for running semi-trusted code package installs, build scripts, CI jobs, unfamiliar repos with controlled side effects, and it can also compleme...

agent-vault v0.39.3

HTTP credential proxy and vault An open-source credential broker by Infisical that sits between your agents and the APIs they call. Agents should not possess credentials. Agent Vault eliminates credential exfiltration risk with brokered access. New here? Thelaunch blog post has the full story...

Prediction Market Betting Is Getting People Banned and Arrested

This week on Uncanny Valley, we dig into the latest prediction market buzz, Flock’s AI-powered police search tool, and how tech bros don’t know how to talk about “rouge” AI agents...

CVE-2026-78583 Incorrect Authorization in Kibana Leading to Privilege Escalation

Incorrect Authorization CWE-863 in Kibana can lead to privilege escalation via Input Data Manipulation CAPEC-153. Elasticsearch cluster privilege declarations originating from integration packages were not validated before being used to mint credentials for enrolled Elastic Agents. A user holding...

pilotprotocol v1.13.10-rc.1

Pilot Protocol The network stack for AI agents. Addresses. Ports. Tunnels. Encryption. Trust. Docs · Wire Spec · Whitepaper · IETF Draft · Agent Skills · Polo Live Dashboard The internet was built for humans. AI agents have no address, no identity, no way to be reached. Pilot Protocol is an overl...

useragent v0.0.109

useragent A comprehensive and categorized collection of User Agents. Installation Instructions To install the useragent tool, use the following command: go install -v github.com/projectdiscovery/useragent/cmd/ua@latest Usage To display help for the tool, use the following command: ua -h This will...

prismor v1.44.0

Prismor Runtime security for Claude Code, Codex, Langchain and other AI agents frameworks/harness. Prismor is highly customizable based on user's own policies. Observe or enforce mode to see agent activity in a local self-serve dashboard Website • Onboard with Skill...

ActBench v0.1.0

ActBench ActBench is a self-evolving benchmark of behavioral safety in cowork agents. It defines behavioral safety as whether an agent's execution remains within the permissions and state changes required by a benign task, and evaluates realized behavioral risk from execution trajectories rather...

PT-2026-85115

Subscriber Broken Access Control in Agentimus – AI SEO, llms.txt & MCP for AI Agents = 1.51.0 versions...

Malicious .git Configs Can Make Claude, Codex, Cursor, and Other AI Agents Run Attacker Code

Manifold Security has disclosed eight security flaws across seven command-line AI coding agents in which a repository's own Git configuration names a command that the agent runs on the developer's machine, four of them still unpatched at publication. The command executes as the user, outside the...

bromure agentic-coding-v4.6.0

Bromure Secure, ephemeral computing in disposable Linux VMs on macOS. → Full details, screenshots, and downloads at bromure.io This repo ships two sibling apps, both built on Apple's Virtualization.framework: Bromure — every browser session runs in a throwaway Linux VM. Close the window, the VM i...

PT-2026-84855

Name of the Vulnerable Software and Affected Versions Hermes version 0.21.0 Description An issue exists where coding agents execute git status or git diff for context gathering before trust prompts are presented to the user. If a repository is delivered as files such as via zip, sync, or USB rath...