Loading video...

Video Failed to Load

Go Home

Coding agent hacking series 2/3: Codex CLI. It looks seriously secure: sandboxing by default, built in Rust, reviewed by top LLMs from OpenAI. But in our latest demo, one web fetch can chain multiple vulnerabilities from prompt injection to unauthorized command execution outside the sandbox in one shot!

DarkNavyOrg's profile picture

DARKNAVY

5,654 subscribers

10,185 views • 5 months ago •via X (Twitter)

LIVE

Anya Rossi• Live Now

Private livecam show

1 Comments

TravelerOfCode's profile picture
TravelerOfCode5 months ago

@OpenAI sandboxing by default still wont save you if the model chains web fetches. data exfil through agent tool calls is the next big class of bugs

Related Videos

Coding agent hacking series 1/3: Claude Code. Anthropic is building impressively powerful cyber models like Mythos. However, their core coding product can still stumble on security boundaries beyond prompt injection. Our demo shows how web content exploring can be chained with other vulnerabilities to bypass permission checks and execute attacker's commands without your approval ;)
0:37

Coding agent hacking series 1/3: Claude Code. Anthropic is building impressively powerful cyber models like Mythos. However, their core coding product can still stumble on security boundaries beyond prompt injection. Our demo shows how web content exploring can be chained with other vulnerabilities to bypass permission checks and execute attacker's commands without your approval ;)

DARKNAVY

10,125 views • 5 months ago

Hermes can now command an army of coding agents. One prompt, multiple agents spawned. Multiple PRs land all in parallel. Agent Orchestrator now Available for Hermes (by Nous Research)
1:40

Hermes can now command an army of coding agents. One prompt, multiple agents spawned. Multiple PRs land all in parallel. Agent Orchestrator now Available for Hermes (by Nous Research)

Agent Orchestrator

44,642 views • 6 months ago

Web search is now enabled by default for the Codex CLI and IDE Extension 🎉 By default it will use a web search cache but you can toggle live results or if you use --yolo live results are enabled by default. More details in the changelog 👇
0:25

Web search is now enabled by default for the Codex CLI and IDE Extension 🎉 By default it will use a web search cache but you can toggle live results or if you use --yolo live results are enabled by default. More details in the changelog 👇

dominik kundel

265,718 views • 9 months ago

Every #GeminiAtWork update in 2 minutes: 1. One Cloud Agent: Shared memory across CLI, web, mobile & Workspace 2. Trusted Context: Knowledge Catalog & Borderless Lakehouse 3. Built-In Control: Agent Sandbox, Identity, spend caps & Smart Routing
2:33

Every #GeminiAtWork update in 2 minutes: 1. One Cloud Agent: Shared memory across CLI, web, mobile & Workspace 2. Trusted Context: Knowledge Catalog & Borderless Lakehouse 3. Built-In Control: Agent Sandbox, Identity, spend caps & Smart Routing

Google Cloud

18,418 views • 19 days ago

Clay is now available in Codex. Sales Reps can prioritize accounts, research buyers, draft outreach, and send emails — all from OpenAI's coding agent, powered by the Functions your team already built in Clay. It was great working with the OpenAI Developers team on this!
0:32

Clay is now available in Codex. Sales Reps can prioritize accounts, research buyers, draft outreach, and send emails — all from OpenAI's coding agent, powered by the Functions your team already built in Clay. It was great working with the OpenAI Developers team on this!

Jeff Barg

42,623 views • 5 months ago

We build a Skill.md for it🔥 OpenAI Developers #Codex + #GPT-5.5 + #Hyper3D Rodin: one-click animated game assets. (This track looks goofy 😂 Try our Skill in any Coding Agent, interactive #3D Asset from image/text ⬇️
0:30

We build a Skill.md for it🔥 OpenAI Developers #Codex + #GPT-5.5 + #Hyper3D Rodin: one-click animated game assets. (This track looks goofy 😂 Try our Skill in any Coding Agent, interactive #3D Asset from image/text ⬇️

Hyper3D by Deemos

23,641 views • 6 months ago

You can now use TinyFish in Grok Build. Search, Fetch, Agent, and Browser all under one plugin inside SpaceXAI's coding agent. Give Grok Build a real connection to the live web and let it work across any site from your terminal.
0:39

You can now use TinyFish in Grok Build. Search, Fetch, Agent, and Browser all under one plugin inside SpaceXAI's coding agent. Give Grok Build a real connection to the live web and let it work across any site from your terminal.

TinyFish

228,566 views • 3 months ago

Introducing jevgrep - a research agent CLI powered by jev from TypeSafe AI that reduces your coding agent cost by 40% (verified on SWE-bench) Make sure to use the built in skill so your coding agent knows to use jg for context collection
0:30

Introducing jevgrep - a research agent CLI powered by jev from TypeSafe AI that reduces your coding agent cost by 40% (verified on SWE-bench) Make sure to use the built in skill so your coding agent knows to use jg for context collection

David

455,892 views • 1 month ago

Codex CLI will have MULTI-AGENT support in 0.123.0 Here's how it will work: → 1. Add `multi_agent=true` in ~/codex/config.toml under [features] → 2. Toggle with `/experimental` → 3. Three built-in agents (default, explorer, worker) → 4. Define your own agents ``` [agents.researcher] description = "Research-focused role." config_file = "~/.codex/agents/researcher.toml" model = "gpt-5.3-codex" model_reasoning_effort = "high" web_search = "live" sandbox_mode = "read-only" developer_instructions = """ You are the researcher agent. - Gather evidence from primary sources. - Return links, findings, and uncertainty. - Do not edit files. """ ``` → 5. Run more agent threads per session (default: 6) ``` [agents] max_threads = 6 Notes: → OAI released 0.122.0 - but rolled it back! → I built it from source (so you don't have)
0:21

Codex CLI will have MULTI-AGENT support in 0.123.0 Here's how it will work: → 1. Add `multi_agent=true` in ~/codex/config.toml under [features] → 2. Toggle with `/experimental` → 3. Three built-in agents (default, explorer, worker) → 4. Define your own agents ``` [agents.researcher] description = "Research-focused role." config_file = "~/.codex/agents/researcher.toml" model = "gpt-5.3-codex" model_reasoning_effort = "high" web_search = "live" sandbox_mode = "read-only" developer_instructions = """ You are the researcher agent. - Gather evidence from primary sources. - Return links, findings, and uncertainty. - Do not edit files. """ ``` → 5. Run more agent threads per session (default: 6) ``` [agents] max_threads = 6 Notes: → OAI released 0.122.0 - but rolled it back! → I built it from source (so you don't have)

Chong-U

36,205 views • 8 months ago

I liked the codex chrome extension so I cloned it to use in Google Antigravity operates via CLI and can be used by any coding agent to operate your browser :) i've finally automated filing expenses in concur
0:46

I liked the codex chrome extension so I cloned it to use in Google Antigravity operates via CLI and can be used by any coding agent to operate your browser :) i've finally automated filing expenses in concur

Ivan Leo

44,868 views • 1 month ago

7 days until Prompt-to-DeFi launch - Unveiling the first full demo. For the first time, DeFi expertise becomes executable infrastructure. From sentences to execution. Zero coding. All non-custodial. Powered by INFINIT's AI Agent Swarm. Watch Prompt-to-DeFi turn ideas to execution↓
0:29

7 days until Prompt-to-DeFi launch - Unveiling the first full demo. For the first time, DeFi expertise becomes executable infrastructure. From sentences to execution. Zero coding. All non-custodial. Powered by INFINIT's AI Agent Swarm. Watch Prompt-to-DeFi turn ideas to execution↓

INFINIT

133,095 views • 10 months ago

You wanted shadcn — so we built the heinous infra to make it work. Introducing Agent Arena, starting with the best: Claude Code by Anthropic Codex 5 by OpenAI Cursor Agent by Cursor Devin by Cognition Gemini CLI by Google DeepMind Grok Code Fast 1 by xAI 1/ Your prompt launches a containerized agent inside a fresh GitHub repo 2/ The agent can install packages, make tool calls, and write multi-file code 3/ The repo is deployed to Vercel *by the agent*, clonable to your heart's desire
0:25

You wanted shadcn — so we built the heinous infra to make it work. Introducing Agent Arena, starting with the best: Claude Code by Anthropic Codex 5 by OpenAI Cursor Agent by Cursor Devin by Cognition Gemini CLI by Google DeepMind Grok Code Fast 1 by xAI 1/ Your prompt launches a containerized agent inside a fresh GitHub repo 2/ The agent can install packages, make tool calls, and write multi-file code 3/ The repo is deployed to Vercel *by the agent*, clonable to your heart's desire

Grace Li

169,090 views • 1 year ago

I have big news! 📢 OpenAI uses Ratatui for their Rust implementation of codex! 🔥🦀🎉 🌀 codex-rs — Lightweight coding agent that runs in your terminal. ⭐ GitHub: #rustlang #ratatui #tui #openai #ai #llm #terminal #agent #coding
0:29

I have big news! 📢 OpenAI uses Ratatui for their Rust implementation of codex! 🔥🦀🎉 🌀 codex-rs — Lightweight coding agent that runs in your terminal. ⭐ GitHub: #rustlang #ratatui #tui #openai #ai #llm #terminal #agent #coding

Orhun Parmaksız 👾

15,526 views • 1 year ago

I built 🐨 Kody to address all the frustrations I had with personal assistants. One of the best parts about Kody is it augments your favorite agent and encourages it to move from one-off prompt to deterministic triggerable code run in a sandbox.
22:09

I built 🐨 Kody to address all the frustrations I had with personal assistants. One of the best parts about Kody is it augments your favorite agent and encourages it to move from one-off prompt to deterministic triggerable code run in a sandbox.

Kent C. Dodds 🐨

21,716 views • 4 months ago

We stopped asking "which AI coding agent is best" We built one command that picks for you. /auto automatically delegates tasks to Claude Code, Codex CLI, Gemini CLI, Blackbox. The right agent for the right task based on what 30M+ developers actually use. Best result. One place. Nothing else comes close.
1:08

We stopped asking "which AI coding agent is best" We built one command that picks for you. /auto automatically delegates tasks to Claude Code, Codex CLI, Gemini CLI, Blackbox. The right agent for the right task based on what 30M+ developers actually use. Best result. One place. Nothing else comes close.

BLACKBOX AI

36,249 views • 6 months ago

It is finally here: Tidewave now supports Claude Code and OpenAI Codex. Tidewave unlocks the full-stack potential of your favorite coding agent by tightly integrating it with your web app and web framework at every layer, from UI to database. More info 👇
1:13

It is finally here: Tidewave now supports Claude Code and OpenAI Codex. Tidewave unlocks the full-stack potential of your favorite coding agent by tightly integrating it with your web app and web framework at every layer, from UI to database. More info 👇

José Valim

100,739 views • 11 months ago

OpenAI launched Agents in ChatGPT today. It’s great for consumers who want to automate tasks in ChatGPT. But what if you want to build features powered by the same tools? I dropped put same demo prompt into Director, our web agent builder, to generate a script for that. Director enables anyone to build software powered by web agents. You prompt it once, and get a deterministic web agent that you can integrate into your application. It's a great starting point for building powerful buttons that do work FOR your users.
1:05

OpenAI launched Agents in ChatGPT today. It’s great for consumers who want to automate tasks in ChatGPT. But what if you want to build features powered by the same tools? I dropped put same demo prompt into Director, our web agent builder, to generate a script for that. Director enables anyone to build software powered by web agents. You prompt it once, and get a deterministic web agent that you can integrate into your application. It's a great starting point for building powerful buttons that do work FOR your users.

Paul Klein IV

32,299 views • 1 year ago

Ship a new eve agent in one minute. From 1. Add a prompt 2. Pick models & MCP connections 3. Deploy Your agent is in production, ready to chat, backed by a Git repo you own.
0:24

Ship a new eve agent in one minute. From 1. Add a prompt 2. Pick models & MCP connections 3. Deploy Your agent is in production, ready to chat, backed by a Git repo you own.

Vercel

142,982 views • 2 months ago