The argument in one line.
The real power of a new Claude release lives in the harness updates not the model benchmarks, and Dynamic Workflows plus UltraCode are the two Claude Code changes that redefine what a single session can accomplish.
Read if. Skip if.
READ IF YOU ARE…
- You use Claude Code on a Max, Team, or Enterprise plan and want to understand when multi-agent workflows justify the token cost.
- You are building or auditing complex codebases and want a practitioner demo of what 96 parallel sub-agents looks like in real use.
- You want a critical take on how Anthropic communicates rate limits and what the percentage figures actually mean.
SKIP IF…
- You are on a free or Pro plan as Dynamic Workflows and UltraCode require Max, Team, or Enterprise access.
- You came for a deep model benchmark breakdown as the creator explicitly dismisses benchmark coverage as table-stakes noise.
The full version, fast.
Anthropic launched Claude Opus 4.8 and buried the two most significant updates in a secondary announcement section: Dynamic Workflows, which lets Claude Code spin up hundreds of parallel sub-agents for tasks too large for a single pass, and UltraCode, a new effort tier that grants the model autonomy to decide whether to trigger a workflow. Two live demos show a 13-agent ecommerce SEO audit completing in five minutes and a 96-agent codebase bug audit producing a ranked HTML report, together consuming 4% of a Max plan weekly token limit. The core practical lesson is that effort level is now a real architectural decision and these modes are powerful but expensive.
Free for members
Chat with this breakdown — free.
Sign in and you get 23 free chat messages on us — ask for the hook, quote a framework, find the exact transcript moment, generate a markdown action plan. Bring your own key when you want unlimited.
Create a free account →Where the time goes.
00:00 – 00:51
01 · Hook + Benchmark Dismissal
Opens on the Anthropic announcement page; dismisses benchmarks as directional-only; argues harness updates are where the real power lives.
00:51 – 01:49
02 · Dynamic Workflows Explained
Walks through the Anthropic blog post explaining orchestrator and sub-agent architecture, then introduces the UltraCode effort setting.
01:49 – 02:40
03 · Demo Setup: Ecommerce Audit
Shows the VS Code effort toggle turning purple and the terminal rainbow animation for /workflows. Sets up a three-site DTC brand audit prompt.
02:40 – 05:29
04 · Demo 1: 13-Agent Audit Running
Fires the dynamic workflow, watches it fan out 9 audit agents plus synthesis, checks /workflows for phase status, waits for deliverables.
05:29 – 06:45
05 · Results + Design Iteration
Reviews the Gymshark (AU) report scored 88/100 on technical SEO, notes vanilla white-paper design, applies a design system to reformat.
06:45 – 08:27
06 · UltraCode Setup + Bug Audit
Switches effort to UltraCode via /effort menu, sets up a bug audit of the creator's personal Rubric dashboard app.
08:27 – 09:25
07 · 96-Agent Verification Step
UltraCode does a pre-assessment, then fans out 8 auditors; verification layer adds 88 more agents; /workflows shows all phases live.
09:25 – 10:11
08 · Bug Audit Results
HTML report delivered: 1 critical, 15 high, 30 medium, 24 low, totaling 70 unique bugs each independently verified.
10:11 – 11:22
09 · Token Cost and Rate Limit Critique
Account usage jumped from 2% to 6% weekly. Argues Anthropic should display absolute token counts instead of percentages.
Lines worth screenshotting.
- Anthropic buried the most significant Opus 4.8 updates below the benchmark table most viewers never scroll past.
- Dynamic Workflows lets a single Claude Code session orchestrate hundreds of parallel sub-agents, turning multi-day agency work into a five-minute deliverable.
- UltraCode is not just a higher effort setting; it gives the model autonomy to decide on its own whether a task warrants spawning sub-agents.
- The /workflows slash command provides a live status panel showing every agent phase, token count, and completion percentage for long-running tasks.
- UltraCode ran a pre-assessment pass before fanning out: 8 auditors first, then 88 independent verification agents to adversarially confirm each finding.
- Two heavy demos consumed 4% of a weekly Max plan token budget, making effort level a genuine cost variable.
- A technically correct report that ships as an unstyled white paper will not be read; prompting for a design system on the second pass is part of the workflow.
- The model improvement from Opus 4.7 to 4.8 is incremental; the harness improvements are the compounding differentiators.
- Dynamic Workflows are available in research preview on Claude Code CLI, Desktop, and VS Code for Max, Team, and Enterprise plans.
- Anthropic using percentage-based rate limit display makes it structurally impossible to verify whether advertised rate-limit increases are meaningful.
Two Claude Code modes that change how you scope a task
WHAT TO LEARN
Dynamic Workflows and UltraCode raise the ceiling on what a single session can accomplish, but they introduce real token costs that make effort level a genuine architectural decision rather than a preference.
- Dynamic Workflows fans a session out into hundreds of parallel agents, turning multi-day research or audit work into a five-minute deliverable, but token cost scales proportionally.
- UltraCode gives the model autonomy to decide whether a task warrants spawning sub-agents, so complex prompts get orchestration automatically without requiring an explicit workflow invocation.
- The /workflows slash command shows a live status panel with agent count, phase completion, and per-agent token usage so you can monitor a long-running task without waiting blind.
- A technically complete output that ships as an unstyled document will not be read; prompting a design pass as a second step is part of the workflow, not an afterthought.
- Two heavy demos consumed 4% of a weekly Max plan limit, making effort-tier selection a real cost variable; reserve UltraCode for tasks where correctness across many files justifies the spend.
- Anthropic presenting rate limits as a percentage of weekly budget makes it structurally impossible to verify whether rate-limit increases are meaningful; absolute token counts would give users real transparency.
Terms worth knowing.
- Dynamic Workflows
- A Claude Code capability that allows one orchestrator agent to spawn and manage hundreds of parallel sub-agents within a single session, tackling tasks too large or complex for single-pass completion.
- UltraCode
- A Claude Code effort setting above xHigh that combines maximum reasoning depth with model-level autonomy to decide when a task warrants triggering a dynamic workflow automatically.
- Orchestrator agent
- The top-level agent in a dynamic workflow that plans phases, assigns sub-tasks to parallel agents, monitors progress, and synthesizes all sub-agent outputs into a final deliverable.
- Adversarial per-finding verification
- A verification strategy where each individual finding from a primary audit is independently re-verified by a separate agent, increasing confidence that reported issues are real rather than hallucinated.
- /workflows
- A Claude Code slash command that displays a real-time status panel for any running dynamic workflow, showing agent phases, individual agent token usage, and completion status.
- Effort control
- A Claude Code setting accessible via /effort that controls how many reasoning passes the model makes before responding, ranging from low to max to UltraCode.
Things they pointed at.
02:40productRoboNuggets community
06:47productRubric personal dashboard app
Lines you could clip.
00:00
“While most people are focusing on these trust me, bro, benchmarks, I think the real unlock are the major improvements that they made to their coding harness.”
04:05
“It costed us 4% of our weekly rate limit. So that tells you, number one, how token intensive these modes are, so be warned.”
10:11
“I would much rather them have an absolute number of tokens in here so that we can really measure when they're saying that they're increasing rate limits or not.”
Word for word.
metaphor
00:00So Cloud Opus four dot eight is here. And while most people are focusing on these , I think the real unlock are the major improvements that they made to their coding hardness Cloud Code. So in this video, I'll give you a rundown of this ultra code mode and how to use it and also their new dynamic workflows feature that is Anthropic's answer to long running agents.
00:16Let's dive into it.
00:21So OPUS four dot eight is now out. And as usual, they published the benchmarks in here, which you can just browse on your own time. But at least after having worked with these models quite a lot already, I think these benchmarks are good to show incremental improvements on where these new models and releases are directionally strong at.
00:36But in my experience, a lot of the power of these models now don't actually directly come from the models, but mostly come from the updates to the harness itself. So if you scroll down to their update in here, they sort of buried the lead in here to put this dynamic workflows feature in this also launching today section.
00:53And if you go through their document, dynamic workflows in Cloud Code, basically, what it does is it helps Cloud take on the most challenging task end to end. So what that means is, let's say, have a problem that is too big for one pass by a single agent and you actually want multiple concurrent agents or sub agents to accomplish that task, you can use dynamic workflows in order to have one orchestrator agent, which is gonna be OPUS four dot eight, and have it orchestrate this multiphase plan to accomplish that one big task that you assign to it.
01:22And apart from dynamic workflows, another thing that they released that they, again, sort of buried the lead in is this new Cloud Code specific setting called Ultra Code. And what it does is set the effort level to extra high while also letting Cloud decide automatically when to use a dynamic workflow to handle your task.
01:39So what does that look like in practice, and how can we invoke these? Well, let's just go into a demo and try it out ourselves. So to use dynamic workflows and UltraCode, you can actually use the Versus Code extension.
01:49So if I go to the effort toggle in here once you update your Cloud Code, you can now see that there is this ultra code option which turns it to purple. And, of course, you can also access it via the terminal view. And I'll just demo it here because at least in the terminal view, it seems like the Anthropic team deemed this release so important that they even assigned, like, a custom sort of a rainbow color whenever you type in workflows into your prompt in here.
02:11And then similarly, if you change the effort to ultra code, that also has that nice animation that they even coded just for that release. For our test prompt, what I'm going to do is give it a sufficiently complex task where I'm asking it to use dynamic workflows to audit three ecommerce websites for a direct to consumer growth agency in Sydney.
02:31So it's going to be a brand audit of these three websites. And for each of these domains, we need a technical SEO scorecard, a content and keyword gap analysis, conversion and user experience flags, three quick wins, three mid effort wins, and basically a brand audit that would have taken a mid class agency a couple of days to put together in the past.
02:51So we'll fire that off using dynamic workflows, and we'll actually see what Cloud Code is going to do with this complex task. And just to show you how much tokens this will consume, I'm currently on the max plan for this account. And right now, my weekly rate limits is consumed at the 2% mark.
03:04So we'll see by the end of this test how much percentage tokens that will consume. And by the way, if you're interested in going from just using AI to getting paid for it, then check out the Robo Nuggets community down in the description. We've got founders in there who landed their first client in weeks, live build sessions where we create this stuff together, and the actual templates behind what I just showed in this video.
03:22The community is also the reason these lessons get made, so see that below if that's for you. Okay. So now that it is running, you can see that what it did here is it's starting to fan out nine audit agents in order to do this task for us.
03:34Now what's good about it is that it actually recognize that dynamic workflows is going to be token intensive. So if you are saving up on tokens, this is probably not something that you would want to just do randomly. But at least for the sake of this demo, let's just go ahead and run it.
03:49Now it's saying that the workflow is running in the background with 13 live fetch agents. And interestingly, you can see what it's doing here that it's acting as sort of the manager or the orchestrator of this whole task. So it's saying that it's using the weight productively and pre building the report generator so that the moment the data lands, it can turn it into the deliverables fast.
04:08So that is what I'm talking about with regard to the harness. Yes. OPUS 4 Dot 7 to OPUS 4 Dot 8 is a nice job, and they'll always show good benchmarks whenever these new models release.
04:17But this sort of user experience and the way these agents are architected really matter a lot more versus the benchmarks that you usually see just the front loaded in a lot of these YouTube videos. Alright.
04:28So this has been running for around five minutes now. One thing you can do actually, because it's saying here that 12 out of 13 agents are done, is you can type in slash workflows. And what that now shows is a proper plan that your Orchestrator agent has drafted up around this audit.
04:45So you can see for phase one, which is the audit itself, those nine agents are already done, and it also shows the amount of tokens that they consumed if you're particular about that. There is a phase two around planning, which I assume is basically planning out its output. And then now we have this synthesis agent in order to put all of those learnings together.
05:05So if it's been running for a while and you need a view of how it's going, then this is one way for you to monitor or observe the progress of your long running task. Alright. So now it's done, and it gave us a couple of deliverables.
05:17It gave us three brand reports for each of those websites. It gave us a comparison sheet as well as the summary document. So if you look at the executive summary, you can see it has the ranking of those different ecommerce website.
05:29It has the top three takeaways. So it highlighted here, let's say, the biggest SEO upside, so definitely agree that that should be top of line for the headlines. And if you look at the individual PDF reports, you can see a more summarized version per brand around their technical SEO capabilities, content and keywords, and conversion and UX with a lot more detail down the line.
05:47Now something that I think it should have done or maybe I should have included in the prompt is a proper design look of this report. Because right now, it's very vanilla white paper. No one's really gonna read this type of design.
05:58Right? So what you can do here is to just ask Cloud Code to iterate on this. And at least from my side, I have this robo group design system to update the executive summary and those PDFs so that it is a bit more beautifully designed versus this white paper report that is technically rich but is probably not as enticing to read.
06:16Once that's done, it's now been properly formatted and is just much nicer to read. And from here, you can just tweak all the details that it got, turn them into slides, ask Cloud Code to revise the wording of it as you would usually do. But at least a lot of the hard work and a lot of the research behind this report has already been done by your multiple agents, and it only took, like, five minutes.
06:36K. Now let's do another test. And for this one, let's try out the UltraCode effort.
06:41So if you just type in effort, you can change the effort level here to the UltraCode smarter level. And just to simplify this, if in case you haven't used or tweaked effort before, effort basically pertains to how many times a model thinks about its output before serving it to you.
06:56And for UltraCode, can see the sub headline here. That's essentially extra high, plus it gives the model the ability to decide if it wants to use dynamic workflows or not. So let's just select that.
07:07So now we're in UltraCode. And for this task, I'm giving it this prompt where I'm asking it to audit the Rubrik app. And in case you're new, the Rubrik app is my own personal command center, which I also share with my community.
07:18And, basically, whenever I create, like, micro apps that help me with our work, I just put them here to centralize all of those applications in one view. And so what we'll ask OPUS four dot eight to do is audit that whole vibe coded application and search thoroughly for bugs.
07:33So I'm going to ask for a ranked bug list, and I want a stand alone report that will just provide me a nice view of what the bugs are so that we can fix it later. So I'll fire that off, and I just made it a bit more open ended versus the other tests that we did.
07:47Because I think with UltraCode, what should now happen is that if it deems this task to be large enough, which I think it is, it should go ahead and use dynamic workflows on its own accord and actually fan out those sub agents to hit on this task. Alright. So it's starting that task now.
08:02I just like to point out. So you can see here that when you're on the ultra code effort method, what it's doing here is that right now, it is doing the work as a sole agent, but it has that intelligence to do some initial analysis first and an initial discovery before doing or orchestrating the deep audit.
08:19So you can see here it recognized that UltraCode is on. So it'll orchestrate a fan out audit with adversarial per finding verification. So lots of big words for Opus four dot eight.
08:29But basically, what that means is that initially, I'll do a pre assessment first before doing a proper fan out audit. And if it seems like it's a big task, then I'll decide on my own if I need to . And now here you go.
08:44It's now doing the audit across eight parallel auditors. So we can actually type in slash workflows now, and what it's now showing us is the status of that task. So we have the rubric bug audit.
08:55We have these eight agents that are all doing the work, all OPUS four dot eight, and you can see here the status of each in terms of that usage. So I just came back to this, and I just wanted to show before I continue this. When it did its verification step, what it did is spawn 88 parallel sub agents in order to verify its findings in here.
09:14So I guess that claim out there is that you can spawn hundreds of parallel sub agents in order to do this long running task is apparently true. So it can happen. And with just this bug report audit, it was able to spawn 96 total sub agents in here.
09:27Alright. So the UltraCode run is now done, and it gave us this HTML page that gave us a view of some of the bugs across my personal dashboard, which is not surprising because at least for my version of this dashboard, it is just for personal use. You can see how powerful this is.
09:40Right? If you have an application or a dashboard that you are serving to clients, you can just use UltraCode and dynamic workflows in order to spawn several sub agents, 96 sub agents, at least for this case, to find critical high and medium bugs, which now from here, you can either read through it or just have Claude code address the top ones.
10:00Alright. And one last thing, if we check back on our account and usage, you can see our weekly rate limits jump from two to 6%. So those two tasks, they're very heavy tasks.
10:10It costed us 4% of our weekly rate limit. So that tells you, number one, how token intensive these modes are, so be warned.
10:18And number two, it sort of kind of tells you how token constrained Tropic still is. In my view, this standard of presenting our rate limits as a percentage should actually be changed. Like, I would much rather them have, an absolute number of tokens in here so that we can really measure when they're saying that they're increasing rate limits or not so that it's much more transparent.
10:36Similar to how you're accessing the Internet and you exactly know how much data you're using as part of your mobile data plans, for example. But anyway, that's a topic for another video. But there you go.
10:45That is UltraCode and dynamic workflows. OPUS four dot eight is great. It's a great incremental release.
10:51But I think for most use cases, the way we work is probably going to be dictated more by the updates that they do to their harness. And UltraCode and dynamic workflows are just two updates that I think are worth paying attention to. If that's useful, then consider subscribing because that helps us a lot to put out more educational content like this.
11:07As always, thanks for sticking until the end, and I'll see you guys next time. Thank you.
The Hook
The bait, then the rug-pull.
The title buries the thesis before the video even starts: the model is not the story. A solo demo creator opens on Anthropic's benchmark table, waves it off as directional noise, then scrolls to the section most viewers would never reach.
Named ideas worth stealing.
00:51model
Dynamic Workflow Architecture
- Orchestrator agent (Opus 4.8)
- Parallel sub-agents per phase
- Synthesis agent
- /workflows monitoring layer
A four-layer structure where one orchestrator plans phases, parallel agents do the work, a synthesis agent consolidates, and /workflows gives real-time visibility.
Steal forAny complex research, audit, or code review task where correctness across many files matters more than token cost.
06:45list
Effort Ladder
- Low
- Medium
- High
- xHigh
- Max
- UltraCode
Six effort tiers in Claude Code; UltraCode adds model-level autonomy to trigger dynamic workflows without explicit user instruction.
Steal forCalibrating session cost vs. correctness for different task types.
05:29concept
Two-Pass Delivery
Run the research or audit agent first for technically correct output, then prompt with a design system to make it client-ready.
Steal forAny client-facing deliverable generated by AI.
How they asked for the click.
VERBAL ASK
11:11subscribe
“If that's useful, then consider subscribing because that helps us a lot to put out more educational content like this.”
Single closing line after all main content, no pre-CTA pitch, low pressure.
MENTIONED ON CAMERA
FROM THE DESCRIPTION
PRIMARY CTAWhere the creator wants you to go next.
AFFILIATECommission earned if you click.
Visual structure at a glance.
