Claude Code and Codex Got an Upgrade Nobody Announced
Mark Kashef reveals that when Claude Code and Codex shipped their record-and-turn-into-a-skill feature, both also quietly gained the ability to process raw MP4 video out of the box, breaking it into frames plus audio and transcript without any special skill or plugin, and he demonstrates three ways to use this: dragging in a video, pointing at a generic link, and hacking the record feature to audit a workflow instead of building a skill.
Mark Kashef13 minTranscript found
Quick learning frame
Read this before watching.
Coding-agent workflow is the loop of inspect, route, plan, edit, verify, summarize, and decide what should be automated next.
New playlist item from Mark Kashef; queued for transcript-backed review, topic mapping, and a practical learning artifact.
Skill you build: The ability to feed raw screen recordings or video links directly to Claude Code or Codex as a substitute for writing out detailed requirements, letting the agent watch, listen to, and interpret a process or design reference frame-by-frame to produce a plan or SOP.
Watch for the shift from claim to mechanism. The learning value is the point where the transcript reveals a repeatable action, tool boundary, context move, review habit, or artifact.
Concept diagram
Where this video fits.
01Inspect context
02Route tool
03Plan work
04Edit safely
05Verify behavior
06Report next step
Deep lesson
Turn this video into working knowledge.
2,808 cleaned transcript words reviewed across 778 timed caption segments.
Thesis
Claude Code and Codex Got an Upgrade Nobody Announced teaches a practical coding-agent workflow move: Mark Kashef reveals that when Claude Code and Codex shipped their record-and-turn-into-a-skill feature, both also quietly gained the ability to process raw MP4 video out of the box, breaking it into frames plus audio and transcript without any special skill or plugin, and he demonstrates three ways to use this: dragging in a video, pointing at a generic link, and hacking the record feature to audit a workflow instead of building a skill.
The goal is not to remember the video. The goal is to extract the operating principle, tie it to timestamped evidence, test how far the claim transfers, and make something reusable.
0:00
Quiet video upgrade
“A few weeks ago, something became infinitely easier to do in both Claude Code and Codex. Both of them shipped a feature that lets you record your screen and turn that recording into a skill. But what most...”
Both Claude Code and Codex can now take a raw MP4 (drag-and-drop, or a generic link like Loom) and break it into second-by-second frames married with the extracted audio and transcript, understanding what was said, how it was said, and what was on screen, a capability that previously required a model like Gemini and was never officially announced by Anthropic or OpenAI. Drag any short screen recording you already have into Claude Code or Codex and ask it to watch end-to-end (not just pull the transcript) to confirm it breaks the video into frames plus audio for you.
4:02
Video as the spec
“underlying codebase, it'll be able to make the changes, make a plan without you saying a single thing outside of this video's context. So, all you need to do is go and download this video, and once you...”
Instead of translating a screen recording into written requirements yourself, you can feed the raw Loom link or MP4 directly to the agent with a prompt like 'watch this end-to-end, don't just pull the transcript, and list everything we need to improve,' and it will produce a full improvement list with exact timestamps tied to your feedback, then build a plan against the codebase. Record a short walkthrough of one thing you want changed in your own project and feed the raw video link straight to your coding agent instead of typing out the requirements list yourself.
9:08
Hacking record-and-replay for SOPs
“be able to start the process of explaining to Codex exactly what I'm trying to do, how I'm looking to build it without having to over-explain and go through 10-20 prompts before we get even close to the...”
By explicitly telling Codex's record-and-replay skill not to build a skill, Mark uses it purely to watch him work through a manual research process (checking X, using Grok for use-case searches) for a set time, then asks it to produce a full SOP identifying steps that could be automated via the X API instead of just recreating the actions as a skill. Next time you catch yourself doing a repetitive manual workflow, record it with the record feature, tell the agent explicitly not to turn it into a skill, and ask instead for an SOP plus automation opportunities.
01
Inspect context
Start with this video's job: Mark Kashef reveals that when Claude Code and Codex shipped their record-and-turn-into-a-skill feature, both also quietly gained the ability to process raw MP4 video out of the box, breaking it into frames plus audio and transcript without any special skill or plugin, and he demonstrates three ways to use this: dragging in a video, pointing at a generic link, and hacking the record feature to audit a workflow instead of building a skill. Treat "Inspect context" as the outcome you are trying to make visible, not a topic label. Anchor it to 0:00, where the video says: “A few weeks ago, something became infinitely easier to do in both Claude Code and Codex. Both of them shipped a feature that lets you record your screen and turn that recording into a skill. But what most...”
02
Route tool
Use "Route tool" to locate the part of the coding-agent workflow mechanism the video is demonstrating. Ask what changes in your real setup if this claim is true. Anchor it to 4:02, where the video says: “underlying codebase, it'll be able to make the changes, make a plan without you saying a single thing outside of this video's context. So, all you need to do is go and download this video, and once you...”
03
Plan work
Turn "Plan work" into the reusable artifact for this lesson: A coding-agent routing and execution matrix with context needed, tool choice, verification, and done signal. This is where watching becomes something you can inspect and reuse.
04
Edit safely
Use "Edit safely" as the application surface. Decide whether the idea touches a browser flow, a local file, a model choice, a source document, a UI, or a review step.
05
Verify behavior
Use "Verify behavior" to prove the lesson. The evidence should connect back to the video title, transcript anchors, and a concrete output, not a generic best-practice claim.
06
Report next step
Use "Report next step" to carry the idea forward: save the prompt, checklist, diagram, or operating rule that would make the next agent run better.
Example
Source-backed artifact packet
Convert the video into a scoped artifact request that includes the transcript claim, mechanism, acceptance criteria, and proof. The output should be a coding-agent routing and execution matrix with context needed, tool choice, verification, and done signal..
Example
Coding-agent workflow proof brief
Separate what the speaker claims, what the demo actually proves, and what still needs outside verification before you adopt the coding-agent workflow pattern.
Example
Teach-back module
Transform the lesson into a definition, a Inspect context -> Route tool -> Plan work -> Edit safely -> Verify behavior -> Report next step diagram, one misconception, one practice exercise, and a check-for-understanding question.
Do not learn it wrong
Treating the title as the lesson without checking what the transcript actually says.
choosing tools by hype
losing context across agents
letting parallel sessions become invisible
Letting the lesson drift into generic Codex vs Claude comparison.
Letting the lesson drift into feature lists without task routing.
Letting the lesson drift into claims that ignore limits or recovery.
Do not count this as learned until these are true.
01
State the transcript-backed claim in your own words: Mark Kashef reveals that when Claude Code and Codex shipped their record-and-turn-into-a-skill feature, both also quietly gained the ability to process raw MP4 video out of the box, breaking it into frames plus audio and transcript without any special skill or plugin, and he demonstrates three ways to use this: dragging in a video, pointing at a generic link, and hacking the record feature to audit a workflow instead of building a skill.
02
Explain the practical stakes without hype: New playlist item from Mark Kashef; queued for transcript-backed review, topic mapping, and a practical learning artifact.
03
Map the idea onto the Inspect context -> Route tool -> Plan work -> Edit safely -> Verify behavior -> Report next step sequence and name the weakest link.
04
Produce the artifact and include the evidence that proves it: A coding-agent routing and execution matrix with context needed, tool choice, verification, and done signal.
Put it into practice
Give this grounded prompt to Codex or Claude after watching.
You are helping me turn one specific YouTube video into real, durable learning.
Source video:
- Title: Claude Code and Codex Got an Upgrade Nobody Announced
- URL: https://www.youtube.com/watch?v=0I-J1aoxYQY
- Topic: Creative Automation
- My current learning frame: Record a short Loom of yourself performing one repetitive task or reviewing one piece of your own product, feed the raw video (not a written description) to Claude Code or Codex, and compare the plan or SOP it produces to what you would have written manually.
- Why this matters: New playlist item from Mark Kashef; queued for transcript-backed review, topic mapping, and a practical learning artifact.
Transcript anchors from this exact video:
- 0:00 / Evidence 1: "A few weeks ago, something became infinitely easier to do in both Claude Code and Codex. Both of them shipped a feature that lets you record your screen and turn that recording into a skill. But what most..."
- 1:54 / Evidence 2: "ago, I had to lean on something like a Gemini because it was one of the only models in the world that could both interpret and create video. But now you can pick up your Codex subscription, your..."
- 4:02 / Evidence 3: "underlying codebase, it'll be able to make the changes, make a plan without you saying a single thing outside of this video's context. So, all you need to do is go and download this video, and once you..."
- 5:40 / Evidence 4: "enjoy the way I break down concepts, and you want to go infinitely deeper on things like Claude Code and Codex, and get exclusive tips and tricks that you'll never see on YouTube, then you want to check..."
- 7:14 / Evidence 5: "and you can just believe it will also work in Cloud Code just fine. I feed it the MP4, I give it the exact same instruction, watch this end-to-end with the audio to get a full picture of..."
- 9:08 / Evidence 6: "be able to start the process of explaining to Codex exactly what I'm trying to do, how I'm looking to build it without having to over-explain and go through 10-20 prompts before we get even close to the..."
- 11:51 / Evidence 7: "of Codex and Claude code in the past 7 days. And then we put it on something like fast mode and we send this over. It might be able to watch this and know that because I'm walking..."
Video-aware target:
- Prompt lane: Coding-agent workflow
- Mechanism to extract: Find the workflow rule that explains when and how to use Codex, Claude Code, browser control, dashboards, or manual review.
- Artifact to produce: A coding-agent routing and execution matrix with context needed, tool choice, verification, and done signal.
- Artifact must include: task class; agent/tool choice; context packet; verification step; handoff/recovery rule
Your task:
1. Use the transcript anchors above as the primary source packet. If you add outside context, label it clearly as outside context and keep it secondary.
2. Create a source-check table with columns: timestamp, claim, transcript support, what the demo proves, confidence, and what still needs verification.
3. Extract the actual teachable mechanism from the video: Find the workflow rule that explains when and how to use Codex, Claude Code, browser control, dashboards, or manual review. Do not invent claims that are not supported by the title, lesson frame, or transcript anchors.
4. Build a reusable learning artifact: A coding-agent routing and execution matrix with context needed, tool choice, verification, and done signal.
5. Include:
- a plain-English definition of the core idea
- a diagram or structured model using this sequence: Inspect context -> Route tool -> Plan work -> Edit safely -> Verify behavior -> Report next step
- answers to these source questions: What workflow pain is being solved? | What exact coordination mechanism is shown? | What evidence proves it changes the work?
- 3 concrete examples that apply the video idea to real agentic work, such as rate-limit routing; browser verification after a UI edit; long-running /goal session review
- 2 failure modes the video helps prevent, chosen from the transcript evidence and these likely risks: choosing tools by hype; losing context across agents; letting parallel sessions become invisible
- a checklist for the next real workflow, focused on: routing decision, context portability, verification, handoff summary
- one practical exercise with a clear done signal: Route three recent tasks across Codex, Claude, browser checks, and manual review with a reason for each.
6. Add a "learning transfer" section: what changes in my workflow tomorrow if I actually learned this?
7. Add a "source check" section that cites which transcript anchor supports each major takeaway.
Quality bar:
- Make this specific to "Claude Code and Codex Got an Upgrade Nobody Announced", not a generic Creative Automation essay.
- Each workflow rule must point to a timestamped claim or demo moment, then state what remains unproven.
- Prefer operational examples, failure modes, and reusable artifacts over broad definitions.
- Call out uncertainty instead of smoothing over weak evidence.
- Avoid these generic drifts: generic Codex vs Claude comparison; feature lists without task routing; claims that ignore limits or recovery.
- If evidence is weak or missing, stop and say what transcript segment or timestamp needs review instead of guessing.
- Finish with a concise artifact I could paste into my learning app.
Misconceptions
What to stop believing.
Creative AI removes the need for taste.
It increases the need for taste because output volume explodes.
The best prompt is enough.
References, critique, iteration, and post-production matter just as much.
Practice studio
Learning only counts when you make something.
01
Transcript evidence map
Separate what the video actually says from what you already believe about the topic.
3 source-backed takeaways with timestamps, confidence, and a transfer note.02
One useful artifact
Apply the video to a real workflow and produce a coding-agent routing and execution matrix with context needed, tool choice, verification, and done signal..
A reusable artifact with a done signal and one verification step.03
Coding-agent workflow teach-back card
Explain the coding-agent workflow mechanism to someone who has not watched the video yet.
A 90-second explanation, one diagram, one example, and one misconception to avoid.
Recall check
Answer first, then reveal — without rewatching.
What capability did Claude Code and Codex quietly gain when they shipped the record-to-skill feature, and why hadn't people needed to rely on Gemini for this anymore?
How does Mark turn a Loom recording into a website improvement plan without writing out requirements himself?
What explicit instruction does Mark give the record-and-replay skill so it audits his workflow instead of just building a skill from it?
Source shelf
Use the video as a doorway, then verify with primary sources.
