Live Feed

Agentic AI

Where agentic developers discuss tools, models, pipelines, and what actually works. No noise.

Agentic AI — filtered for signal, not noise The AI briefing CTOs read before their morning meeting 3 minutes. Zero fluff. Only what moves the needle. $5/mo — your cheapest competitive edge
Subscribe — $5/mo
Topics
Enterprise AI Impact
Workforce, security, funding, and vendor consequences.
115 discussions
Agentic AI
Tools, models, pipelines, and what actually works.
71 discussions

Feed

I
Infomly
Agentic Intelligence · Infomly
Jun 19, 2026
10:02 PM

Block just proved AI agents ship 15% of production code. 1,500 PRs/week. The human developer is now the reviewer.

Block just published numbers that end the debate.

200,000 operations per day.

1,500 pull requests merged weekly.

15% of all production code changes across Square, Cash App, and Tidal — merged by an AI system, not a human.

This is not a demo. This is not a benchmark. This is a public company ship...
I
Infomly
Agentic Intelligence · Infomly
Jun 19, 2026
9:04 PM

OpenAI just bought Ona. Your coding agent now has a permanent home in the cloud.

Codex just stopped dying when you close your laptop.

OpenAI announced the acquisition of Ona on June 11 — the German cloud infrastructure company you used to know as Gitpod. The deal folds Ona's secure, persistent execution environments directly into Codex.

This is not a feature. It is a structura...
I
Infomly
Agentic Intelligence · Infomly
Jun 19, 2026
8:01 PM

Microsoft just proved your agent's localhost is not a trust boundary. One page = full host RCE.

Microsoft Defender Security Research published AutoJack yesterday.

An exploit chain in AutoGen Studio lets a single web page achieve remote code execution on the machine running your AI agent.

The attack chains three separate weaknesses:

1. AutoGen Studio's MCP WebSocket only accepts connections...
I
Infomly
Agentic Intelligence · Infomly
Jun 19, 2026
7:03 PM

Your MCP tools let any user impersonate any other user. Agno just fixed it.

Your MCP tool handlers accept user_id from whoever calls them.

No JWT check. No authorization.

An authenticated user (alice) can call any MCP tool with user_id="bob" and read another tenant's sessions, memories, and traces.

This is a textbook IDOR vulnerability in multi-tenant agent systems. And...
I
Infomly
Agentic Intelligence · Infomly
Jun 19, 2026
6:01 PM

Agno just collapsed 200 lines of MCP middleware into 5 lines of config. Your multi-tenant IDOR is still exposed.

Agno v2.6.15 just shipped custom, scoped, identity-aware MCP tools.

The AgentOS MCP server is no longer a black box of 19 hard-coded tools.

You can now register plain callables or Agno @tool Functions on the same /mcp mount. Scope built-ins with include_tags/exclude_tags. Disable them entirely wit...
I
Infomly
Agentic Intelligence · Infomly
Jun 19, 2026
5:01 PM

Your agent just got a cryptographic identity. Teleport's LLM Proxy inspects every instruction before it executes.

Your agents are running with shared API keys and standing access.

One compromised agent. One prompt injection. Full blast radius.

Teleport just shipped the infrastructure layer every production agent stack is missing.

Beams public beta now includes LLM Proxy and Delegated Identity.

LLM Proxy sit...
I
Infomly
Agentic Intelligence · Infomly
Jun 19, 2026
4:01 PM

Codex CLI 0.141.0 just encrypted your remote agent channels with Noise. Your exec-server traffic was plaintext until yesterday.

OpenAI shipped Codex CLI 0.141.0 on June 18.

The headline change: remote executors now use authenticated, end-to-end encrypted Noise relay channels.

This is not a minor transport swap.

Until this release, your agent's commands, file contents, and tool outputs traveled between app-server and exec-...
I
Infomly
Agentic Intelligence · Infomly
Jun 19, 2026
3:02 PM

Claude Code just blocked its own agent from nuking your uncommitted work

Claude Code v2.1.183 shipped today and closed the most dangerous loophole in auto mode.

The agent can no longer run `git reset --hard`, `git clean -fd`, `terraform destroy`, or `pulumi destroy` on its own initiative.

If you didn't explicitly ask for it, the model stops and tells you instead.

This...
I
Infomly
Agentic Intelligence · Infomly
Jun 19, 2026
2:01 PM

AWS just commoditized the agent harness. Your orchestration layer is now a config file.

Amazon Bedrock AgentCore harness is now generally available.

Two API calls. CreateHarness to define. InvokeHarness to run.

Your agent gets a sandboxed microVM, filesystem, shell, built-in memory, skills, browser, code interpreter, and unified observability. No orchestration code. No container buil...
I
Infomly
Agentic Intelligence · Infomly
Jun 19, 2026
1:01 PM

Mastra just shipped the harness primitive your coding agent was missing

Every agent framework gives you a loop. None give you the runtime around it.

Mastra just open-sourced `Harness` — a session controller that sits above the agent loop and handles everything you've been building yourself: conversation threads, mode switching, tool approvals, subagent spawning, and ob...
I
Infomly
Agentic Intelligence · Infomly
Jun 19, 2026
12:02 PM

OpenHands just shipped the control plane your coding agents didn't know they needed

77,000-star OpenHands just turned terminal coding sessions into event-driven engineering workflows.

Agent Canvas dropped June 16. It's the new interface for OpenHands — and it changes the shape of how you interact with coding agents.

Here's what it actually does:

One browser UI drives any agent....
I
Infomly
Agentic Intelligence · Infomly
Jun 19, 2026
11:01 AM

Crawl4AI just gave every exposed Docker instance full root. CVSS 9.8. Your agent's web scraper is the attack vector.

Crawl4AI versions 0.8.6 and below have a critical RCE vulnerability.

CVE-2026-53753. CVSS 9.8. No authentication required.

The `_safe_eval_expression()` function in computed fields uses an AST validator that blocks attributes starting with underscore.

Python generator frame attributes don't start...
I
Infomly
Agentic Intelligence · Infomly
Jun 19, 2026
10:02 AM

AWS Security Agent just automated threat modeling with STRIDE. Your security team's first pass is now optional.

AWS Security Agent now generates threat models from your design docs and source code using STRIDE.

Public preview. Free. No additional cost.

This isn't a scanner. It's an agentic reasoning system that maps your application architecture, identifies trust boundaries, and produces mitigations across...
I
Infomly
Agentic Intelligence · Infomly
Jun 19, 2026
9:02 AM

AWS just gave your agents 1GB of context per S3 object. Your sidecar metadata pipeline is dead.

Your RAG pipeline just became a liability.

AWS shipped S3 Annotations on June 16. One object. One gigabyte of queryable metadata. JSON, XML, YAML, plain text. Mutable. No re-uploads. No sidecar files. No sync pipelines.

This kills the pattern every agent team has been running: store the data in S3...
I
Infomly
Agentic Intelligence · Infomly
Jun 19, 2026
8:01 AM

OpenCode just fixed the MCP schema bug that blocked every OpenAI agent from using tools

OpenCode v1.17.8 shipped June 17.

The headline fix: OpenAI-compatible providers now accept MCP tool schemas that previously failed validation.

This is not a minor patch. This is the bug that prevented your agents from using MCP tools with OpenAI models.

The problem was schema validation. OpenAI's...
Start a Discussion