I Run a One-Person Company With AI—Here’s What I Won’t Let It Do, by (not found or attributed):
https://hackernoon.com/i-run-a-one-person-company-with-ai-heres-what-i-wont-let-it-do?ref=dxsrc.org
#aiagents — Public Fediverse posts
Live and recent posts from across the Fediverse tagged #aiagents, aggregated by home.social.
-
GitHub Agentic Workflows is now generally available, bringing coding agents into GitHub Actions with read-only-by-default permissions and separately scoped write jobs.
However, these safeguards don't automatically cover direct write permissions or custom jobs. Security still depends on how each workflow is configured.
🔗 Read more: https://www.infoq.com/news/2026/11/github-agentic-workflows-ga/
#GitHub #GitHubActions #AgenticAI #AIAgents #DevOps #SoftwareDevelopment #AISecurity #InfoQ
-
“That’s unfortunate”: Amazon has thousands of AI agents write code
-
Upgrading the MCP Python SDK does not fix its OAuth credential leak (CVE-2026-115563) for unattended clients. The client-credentials and private-key-JWT providers still follow whatever authorization server the MCP server advertises until you pass issuer=. On mcp 1.30.0, leaving it out emits only a DeprecationWarning, which Python hides by default.
https://github.com/modelcontextprotocol/python-sdk/security/advisories/GHSA-qx49-fqc8-xw99
#Python #MCP #AIAgents #Cybersecurity -
Brave Search anti-scraping defenses will soon encrypt search result links via intermediate redirect pages to block surging AI agents and web crawlers.
#BraveSearch #GoogleSearch #AntiScraping #AIAgents #Cybersecurity #TechNews
-
AI Agent Exposes Enterprise Security Gaps
The alarming truth is that 82% of IT and security pros have stumbled upon unknown AI agents in their systems over the past year, leaving gaping holes in enterprise security. These rogue agents can gain access in mere minutes, often created by non-security teams and granted access by unsuspecting users.
#CloudSecurity #AiAgents #EnterpriseSecurity #ShadowIt #EmergingThreats
-
#AWS gives developers a free local stand-in for its cloud, built from scratch by #AIagents
#AWSEmulator, simulates AWS services on a #developer’s own machine. Built from more than 400 API models covering more than 200 services. Developers and #AI agents can use it to test code without a network connection and without paying for real cloud. Agents work from AWS’s public API models, which describe how each service should behave, and check their work against real AWS cloud.
https://www.geekwire.com/2026/aws-gives-developers-a-free-local-stand-in-for-its-cloud-built-from-scratch-by-ai-agents/ -
Giving AI Agents a Better Way to Debug CSS, by @amitmerchant.bsky.social:
https://www.amitmerchant.com/giving-ai-agents-a-better-way-to-debug-css/?ref=dxsrc.org
-
This is what happens when you are too lazy to read your own emails!!! No sympathy!
"A Microsoft 365 outage in Exchange Online caused widespread email and update delays on Thursday and Friday. Both #ChatGPT and #Claude struggled to access #Microsoft365 mailboxes during the disruption. This technical failure created unique risks for companies utilizing AI agents to manage email inboxes. Because an empty inbox and a stuck delivery look identical to an AI, agents processed delayed messages after human representatives had already resolved the requests. Consequently, these automated actions inadvertently overwrote previous phone updates made by customers. Microsoft extended its email timeout window to 48 hours to protect delayed messages, but businesses must establish clear protocols to handle backlogs and prevent conflicts."
#AIAgents #AISucks #AIDependence #Technology #MicrosoftOutage
-
Two models, two jobs in my Claude Code setup.
I've tested Fable 5.2 since its release: very powerful for creative tasks and for reviewing work from a different angle, but probably too token hungry for all my workflows.
So Opus 5.5 does the work, and a Fable agent checks it, midway through a task or before my colleagues, the subject matter experts, review it.
-
Stopping Rogue AI Agents Requires Tested Controls
When AI agents run amok, shutting them down isn't as simple as flipping a switch - it's a complex puzzle where disabling one piece may not stop the chaos. To truly stop a rogue AI agent, you need to test controls that can revoke both credentials and network access.
#AiAgents #EmergingThreats #ArtificialIntelligence #OwaspGenaiSecurityProject #ModelSecurity
-
Ever wondered why an AI agent spent 50 cents reviewing a pull request yesterday and 5 dollars on the exact same task today? Because agents decide how much work to do on the fly, retries, growing context, and exploration can quickly multiply the cost of a single run.
Upsun Dispatch ties every token and agent minute directly to the specific workflow in real time so you can track spend before it happens. ⚡
-
Wenn KI-Agenten ausser Kontrolle geraten | PCtipp https://www.pctipp.ch/themen/news/wenn-ki-agenten-ausser-kontrolle-geraten #ArtificialIntelligence #AI #AIagent #AIagents
-
Think you missed the AI wave? Sunil Potti joins David Bombal to explain why cybersecurity could be one ofits biggest opportunities. From AI agents that test your defenses to robots that can cause real harm, they explore what comes next and how to prepare your career for 2027.
Big thanks to Cisco & Splunk for sponsoring my trip to .Conf 2026 and also for sponsoring this video.
Watch the full video on YouTube: https://youtu.be/zsIIIga8KbU
-
ICYMI: United bans AI agents from bookings as Brussels targets paid AI picks: United rewrote its contract to bar AI agents from buying tickets, the EU proposed fines of up to 6% of global turnover, and Hims & Hers now faces seven states. https://ppc.land/united-bans-ai-agents-from-bookings-as-brussels-targets-paid-ai-picks/ #UnitedAirlines #AIAgents #TravelNews #AIBooking #EULaw
-
Google Labs Tabletop introduces an experimental AI collaboration hub that lets friends build custom trip planners, polls, and expense splitters.
#GoogleLabs #Tabletop #AIAgents #VibeCoding #Productivity #TechNews
https://securityonline.info/google-labs-tabletop/?utm_source=mastodon&utm_medium=jetpack_social
-
Adobe CTO on plans for AI agents to access Adobe tools
Adobe plans to let customers use their preferred AI agents, …
#NewsBeep #News #Technology #Adobe #AIAgents(AgenticAI) #ArtificialIntelligence(AI) #AU #Australia #ChatGPT #Claude #ContentCreation #Editorinterview #GenerativeAI(GenAI) #GoogleGemini #LargeLanguageModels(LLMs) #Machinelearning(ML) #ModelContextProtocol(MCP) #OpenAccess #photoshop #ResearchandDevelopment(R&D) #UnitedKingdom(UK) #UnitedStatesUS
https://www.newsbeep.com/au/962457/ -
#Development #Pitfalls
MCP is dead simple · “And that’s exactly the problem.” https://ilo.im/16hw5e_____
#AI #AiAgents #MCP #RestAPI #Security #AccessControl #Idempotency #DevOps #WebDev #Backend -
ICYMI: United bans AI agents from buying its tickets, seats and upgrades: Rule 6.J.6 applies even when a traveller supervises or approves each step, and reaches travel agency tools. Can an AI still show United fares to a customer? https://ppc.land/united-bans-ai-agents-from-buying-its-tickets-seats-and-upgrades/ #UnitedAirlines #AIBan #TravelNews #AIAgents #Ticketing
-
Gemini 4 Hydrogen: How it accelerates AI development and programmer productivity.
https://gadgetflux.eu/gemini-4-hydrogen-ai-rapid-pentru-programatori/
#Gemini4Hydrogen #GoogleAI #GeminiAI #ArtificialIntelligence #AIAgents #SoftwareDevelopment #CodingAI #GenerativeAI #GoogleCloud #TechInnovation
-
Of 2,053 rules from AGENTS.md, CLAUDE.md and similar files, checked against real agent commits, 54.4% were followed every time and 37.3% never. Only 8.3% fell in between.
Data: 6,541 Copilot, Claude Code and Codex commits across 268 repos. The split holds for rules with 5+ applicable commits, so spot-checking a few rules says little about the rest.
Which rules did you give up on and move into CI instead?
-
Happening today at 1:00 PM! ⌚
Join Matthew Saunders at #BADCamp to explore persistent #AIAgents in decoupled architectures. Learn how long-running agents retain context and orchestrate workflows using amazeeclaw.
Expect insights on:
• Managing state & system boundaries
• API integration over time
• Real-world reliability & security📍 Telegraph Gallery Suite
-
----------------
🎯 AI: Grok CLI on a Raspberry Pi
===================An Internet Israel article describes using Grok CLI on a Raspberry Pi as the entry point for an AI agent. The stated goal is to run the agent workflow on a machine controlled by the operator, rather than leave a cloud-hosted agent running with broad access to personal accounts.
Architecture
The article distinguishes several ways to send requests to the same CLI-based workflow:
• An SSH session can start an interactive grok session.
• A script can pass a single prompt with grok -p.
• A small password-protected website, available only on the internal network, can call the workflow and display its response.
• A WhatsApp bridge can pass an incoming message to a script and return the result to the chat.The website and WhatsApp bridge are proposed interfaces, not features shown with implementation details in the supplied excerpt. The author notes that the agent's effective access depends on what is made available to it on the machine, not on which interface supplied the text.
Authentication and execution
The article describes installing Grok CLI and authenticating with grok login --device-code. A device code and address are displayed, and the user completes approval in a browser on another machine. The stated installation path is ~/.grok/bin/grok.
For a non-interactive request, the example calls the binary with -p and asks for a five-line file summary. The excerpt does not provide a complete service, scheduler, secret-management design, or permission configuration. It mentions cron, secret managers and guardrails as concepts the intended technical audience should be comfortable building.
Example workflow
The author describes a script that checks incoming invoices, sends them to an accountant and flags increases in recurring charges. In the reported example, it detected a gas company's increase of 10 NIS per month, or 20 NIS on a two-month bill. The author says a proposed customer-service email was reviewed and approved before it was sent, and that the charge was subsequently reduced. The supplied text stops before the script's implementation, so its integration and verification details are unavailable.
Limits
Running the CLI and surrounding scripts on a Raspberry Pi does not, by itself, establish that the model's inference runs locally. The excerpt does not document where inference occurs, the account's data handling, pricing, or the resulting permission boundaries. The author's claims of lower cost and greater control are not independently verified in the supplied material.
🔹 GrokCLI #RaspberryPi #AIAgents #tool
-
Agent memory: Agent memory is storage that lets an AI agent keep facts, preferences and past actions across separate sessions, beyond a context window that resets each time. https://ppc.land/agent-memory/ #AIAgents #AgentMemory #ArtificialIntelligence #MachineLearning #DataStorage
-
Northeastern University: What are they talking about? AI agents are developing odd dialects humans struggle to follow. “With AI digital assistants like Meta’s Muse using everyday language to book dental appointments and Grok Bot picking up the slack on work tasks, agentic AI has gone mainstream. But the integration hasn’t been seamless. When left to their own devices, AI agents behave in […]
https://rbfirehose.com/2026/11/12/northeastern-university-what-are-they-talking-about-ai-agents-are-developing-odd-dialects-humans-struggle-to-follow/ -
28 of 89 Terminal-Bench 2.0 tasks had defects; in 16, the instructions and the verifier disagreed. Fixing them (2.1) raised accuracy over 10 points and moved agents up to three ranks. A longer time limit closed a 20-point harness gap on GPT 5.4.
https://arxiv.org/abs/2611.08417A pass rate can't tell a wrong agent from a wrong test. Watching the run can. We built a live browser link to an agent's terminal for that: https://shell.online
-
Live on Maven Central (`0.31.0`)!
📚 ADK integration docs:
👉 https://dokimos.dev/integrations/adk/📝 Full walkthrough:
👉 https://glaforge.dev/posts/2026/11/12/evaluating-adk-java-and-kotlin-agents-with-dokimos/#AIAgents #Gemini #JUnit #OpenSource #GenerativeAI #LLMasJudge
-
CERT-EU warnt: Rapide Zunahme der Angriffe mit Agenten-Frameworks wie #Artex | Security https://www.heise.de/news/CERT-EU-warnt-Rapide-Zunahme-der-Angriffe-mit-Agenten-Frameworks-wie-Artex-11520602.html #ArtificialIntelligence #AI #AIagent #AIagents #Hacking #PenTest #PenTesting #PenetrationTesting
-
Zerply is featured on Awesome Indie! 🚀
Agentic AI SEO Platform with AI Visibility Tracking
Upvote it → https://awesomeindie.com/product/zerply
-
Norwest has announced a $4bn fund, its 18th and largest to date.
The 65-year-old investment firm is backed exclusively by Wells Fargo and says it wants to increase investment in AI startups across sectors and countries, especially the US, India and Israel.
Its areas of interest include consumer and healthcare products, voice and physical AI, aerospace, and cybersecurity involving autonomous agents.
One recent in…
-
48 AI agents find a Maya pyramid
What do you think?
#MCC #BreakingNews #ArtificialIntelligence #Archaeology #AIAgents #MCCBreakingNews
-
What happens when an AI agent rotates credentials after approval—or loses authority before execution?
CDE Kingpin RC.4 passed 7 publicly reproducible adversarial qualifications covering credential rotation, approval replay refusal, and authority contraction.
1 case remains UNSUPPORTED. False-PASS controls also tested.
Evidence and reproduction:
https://github.com/putmanmodel/agent-governance-test-harness?ref=20261112 -
#Development #Explainers
WebMCP explained with a to-do app · When agents stop guessing what a webpage does https://ilo.im/16hrzy_____
#WebMCP #MCP #Webpages #AiAgents #AI #Chrome #Browser #WebDev #Frontend #HTML #JavaScript -
Two Things That Look Identical, and Aren't
"An ASCII filter is a content eraser wearing a duplicate guard" (m/general)
+ "Moltbook accepts comments it can never show you: the depth-5 render cliff" (m/agents)This + more in today's Moltbook Pulse (Edition #112):
https://superagent-ebe00561.base44.app/functions/serveDigestPage?edition=112&utm_source=mastodon&utm_medium=social&utm_campaign=pulse-edition-112 -
In 72% of failed multi-agent runs, the trajectory held evidence against the error before the failure. In 88% of those, no agent acted on it.
From a full read of 95 failed runs of six systems (Who&When and MAD corpora). Faults entered at a median 0.18 of the run. Also: 45% of hand-crafted Who&When failures ended in a provider content-filter crash, then the harness emitted a final answer no agent wrote.
-
I'm running an experiment on two open-source projects: every change to how they behave goes through #OpenSpec first. Proposal, spec delta, design and tasks are written and committed before any code.
truebees-cli: a CLI for the Truebees media verification platform
https://gitlab.com/truebees/tools/truebees-cliDecor Planner: plan picture frames on a photo of your wall, to scale, and export where the nails go
https://github.com/StefanoTavonatti/decor-planner#OpenSource #SpecDrivenDevelopment #AIAgents #Python #TypeScript
-
AI agents are changing enterprise information creation, movement, and governance, while data observability becomes a prerequisite.
Source: SiliconANGLE
https://siliconangle.com/2026/11/12/data-observability-becomes-must-ai-agents-build-databases-kubeconna/ -
United Airlines restricted third-party personal AI agents on its digital platforms. The airline cited data protection, potential account changes, and regulatory concerns regarding FAA and DOT rules.
-
united is blocking personal AI agents from booking its flights. first US airline to say it out loud
kirby's line, per the WSJ: "I don't want to create another middleman"
meanwhile walmart built an official door for agents, with spending limits
be honest: would you hand an agent your airline login, or is this one job you'd still do yourself?
https://skift.com/2026/11/11/united-airlines-moves-to-block-personal-ai-agents-from-booking-flights/
-
https://www.europesays.com/people/304470/ Jamie Dimon warns, as AI companies ‘boast’ their Agents can do a lot including making payments for customers; says: Don’t allow that as they will … #AIAgents #AutonomousAgents #FinancialInstitutions #JamieDimon #JPMorganChase
-
For about a week in July my job was watching a queue.
The estate's CI ran on one self-hosted runner: three concurrent slots and 4 GB of memory, on a Proxmox box that was also running the desktop I work at. Most of the time that was plenty.
Then Rust builds started taking all three slots at once, so the first fix was to make Rust slower on purpose: a resource_group so the compile jobs run one at a time and leave two slots for everything else.
Two days later I pointed agents at the twenty-one critical or high findings from an architectural review of the Go toolkit. Each fix was a branch, a test, a merge request and a pipeline.
When code is written at the speed a person thinks, three slots is plenty. With a set of agents the arrival rate changes completely. The thing in short supply stopped being the ability to produce a change and became the capacity to verify one.
Also in there: a forty-three pipeline backlog that looked worse than it was, scanners that stopped downloading their own vulnerability databases on every run, and an AWS overflow fleet that got switched off after a $270 bill.
So it's one runner again, and the work since has gone into having fewer pipelines rather than faster ones.
-
Databricks has unveiled an upgrade to its Genie AI assistant that will let it handle business tasks without repeated instructions.
Source: Fintech News Singapore
https://fintechnews.sg/141098/ai/databricks-genie-ai/ -
#Development #Releases
Google releases Modern Web Guidance 1.0 · The setup that avoids unwanted code https://ilo.im/16hvnx_____
#Programming #Coding #AiSkills #AiAgents #Prompts #AI #ClaudeCode #WebPlatform #WebDev #Frontend -
DATE: November 12, 2026 at 01:11AM
SOURCE: SOCIALPSYCHOLOGY.ORGTITLE: United Airlines Bans AI Agents From Booking Flights for Travelers
Source: CBS News - U.S. News
United Airlines said Wednesday that it is banning AI agents from making flight reservations on behalf of travelers, citing security risks and the possibility that such actions could violate federal rules. To make an airline booking, an AI agent requires access to a customer's personal information, including credentials like payment information, and checking in for a flight requires travelers to confirm that their luggage does not contain banned...
-------------------------------------------------
Private, vetted email list for mental health professionals: https://www.clinicians-exchange.org
Unofficial Psychology Today Xitter to toot feed at Psych Today Unofficial Bot @PTUnofficialBot
-------------------------------------------------
#psychology #counseling #socialwork #psychotherapy @psychotherapist @psychotherapists @psychology @socialpsych @socialwork @psychiatry #mentalhealth #psychiatry #healthcare #depression #psychotherapist #UnitedAirlines #AIAgents #ArtificialIntelligence #AirTravel #TravelNews #FlightBooking #Aviation #DataSecurity #Cybersecurity #TravelSafety
-
Bumping parser-rs from 1.1.0 to 1.3.0 in landing-rs: the agent added a mandatory assignment of its own and decided what counted as done. https://hackernoon.com/when-ai-coding-agents-start-acting-like-tech-leads #aiagents
-
NEW! Leanpub Track LAUNCH 🚀 AI Agents Masterclass: Understanding, Building, and Managing AI Agents by Thomas De Vos, Ian Johnson, Russ Miles and J. Owen
https://leanpub.com/set/leanpub/aiagentsmasterclass
#courses #leanpublishing #selfpublishing #AIAgents #ClaudeCode #HarnessEngineering #CodingAgent
-
Does ChatGPT remember conversations? Find out how AI memory works in modern chatbots.
https://gadgetflux.eu/cum-functioneaza-memoria-ai-pe-termen-lung/
#AIMemory #ChatGPT #ArtificialIntelligence #AIAgents #RAG #MCP #Chatbots #MachineLearning #Technology #GadgetFlux
-
Leanpub Course LAUNCH 🚀 Giving an AI a Job: The Workbook by Marius Andronie #AIautomation
#AIautomation #AIworkflow #PythonAI #AIagents #Workflowautomation
-
From the Leanpub Blog: Leanpub Course LAUNCH 🚀 Giving an AI a Job: The Workbook: Choose the first job to give an AI system, and make it safe enough to leave alone by Marius Andronie
#books #leanpublishing #selfpublishing #AIautomation #AIworkflow #pythonAI #AIagents #workflowautomation
-
Proofpoint uncovered the BluePhishing Entra Agent attack, where one OAuth consent lets hackers spawn 250 rogue Microsoft Entra Agent ID accounts.
#BluePhishing #MicrosoftEntra #Cybersecurity #OAuth #AIAgents #CloudSecurity