DEV Community: Entire

The #1 Developer Skill in the Agentic Era

Rizèl Scarlett — Tue, 02 Jun 2026 23:12:42 +0000

At first, many dismissed AI as a temporary bubble, but now that engineering organizations are adopting agentic workflows at scale, it is much harder to negate this new reality. Naturally, this shift has triggered a wave of industry anxiety. Engineers are looking at the landscape and asking a frantic question: What skills do I need to keep my job?

For decades, developers prided themselves on memorizing syntax, writing algorithms from scratch, and passing LeetCode interviews by repeating patterns. The pivots we made in the past were typically centered around the framework of the week (especially in the JavaScript world). Switching frameworks only required relying on the strong foundational software engineering skills picked up at school, in a coding bootcamp, or even from a YouTube video. But now that AI agents can produce code in seconds, developer egos are crumbling.

In pursuit of relevance, developers have to sift through the noise of industry influencers pushing various strategies: Agent Skills, MCP, Gas Town, or building custom agents from scratch. Yet every one of these strategies boils down to a single foundation. That same foundation remains the number one skill for developers today, just as it was before the AI boom: communication.

The Value of Communication in Software Engineering

Code alone does not make a great developer. When coupled with strong communication skills, it becomes a distinct competitive advantage. However, we often overlook these skills because our industry excuses poor communication, hiding behind the stereotype that software engineering is an isolated job for people who just want to avoid human interaction.

Early in my career as a junior software engineer, I struggled. It was not due to technical aptitude; instead, I simply struggled to verbalize my thoughts well. For example, when I was stuck on a problem, I didn’t understand how to ask for help. I would write essay-length questions on Slack, burying my actual problem in a mountain of unnecessary context, or I wouldnt provide enough information at all. Either way, my teammates werent motivated to help me because they didn’t understand what I was trying to say.

I finally made progress in my career when I learned how to communicate with clarity. That realization eventually propelled me out of junior loops and into senior leadership roles in developer advocacy, where communication is the core focus.

This requirement is evident in the way our industry identifies talent. In technical interviews, the engineers who get hired are the ones who vocalize their reasoning, explain their trade-offs, and talk through their strategy as they adapt to new constraints. The candidates who pass are rarely the ones who silently type out a perfect linked list.

The Need for Communication in the Agentic Era

// Detect dark theme var iframe = document.getElementById('tweet-2060553984947950017-360'); if (document.body.className.includes('dark-theme')) { iframe.src = "https://platform.twitter.com/embed/Tweet.html?id=2060553984947950017&theme=dark" }

Coding agents have increased the need for clear communication. To have a successful session with an agent, you must be able to clearly express your intent. I find that when engineers claim coding agents are hard to use, unreliable, or hallucinate too much, or when they rely too heavily on overly complex workflows to guide an agent through a codebase, their communication skills often need improvement.

Methods like context engineering, planning, and leveraging Skills are simply different user experiences designed to help engineers better communicate with their agents. For example, while you can create a plan with an agent before starting work, using a command like /plan ensures you provide the right details upfront and saves an artifact of that plan in a markdown file.

Orchestration Makes Communication Less Visible

As agent harnesses and models improve, and as executive leaders demand more output, engineers are getting creative to work faster. One common solution is orchestration. In practice, this means you act as a manager of multiple agents working simultaneously, even while you sleep, and you simply review the output at the end.

AI engineering leaders are starting to point out the trade-offs of this approach. @addyosmani wrote an article about it titled The Orchestration Tax.

// Detect dark theme var iframe = document.getElementById('tweet-2059844244907696186-696'); if (document.body.className.includes('dark-theme')) { iframe.src = "https://platform.twitter.com/embed/Tweet.html?id=2059844244907696186&theme=dark" }

While your agents might produce a massive amount of work, you now have to review a mountain of output. My main issue with this setup is that the underlying reasoning behind the output often gets lost.

Treat Agent Sessions as Engineering Artifacts

// Detect dark theme var iframe = document.getElementById('tweet-2061159078479663601-611'); if (document.body.className.includes('dark-theme')) { iframe.src = "https://platform.twitter.com/embed/Tweet.html?id=2061159078479663601&theme=dark" }

To solve this, we have to start treating agent sessions as durable engineering artifacts. Just like we use version control for our lines of code, we need an audit trail for our agent interactions. You should be able to go back and view the transcripts, the prompts, the tool calls, and the trade-offs an agent made, giving you the context you need to answer for your architectural decisions.

This is one of the many tools my team is building at Entire. We wanted to create that missing track record for agentic workflows. Now, if I open a commit from last week and see a vague message like fixed, I dont have to wonder what happened. I can run a simple command like:

entire checkpoint explain

Entire then surfaces what was asked, what files changed, what the agent reported back, and which commit it belonged to. It is incredibly useful because I can even ask my agent to look at those tracked sessions to help me better understand and expand on the context. This allows me to keep honing my communication skills and progressing as an engineer.

A Real Example

Ill be vulnerable. Yesterday, I published a blog post on our company website. While the tests for the blog post passed in staging, when I merged it to production, it failed to deploy because of a failing Cloudflare test. I worked with my coding agent to diagnose the problem, and then I learned I had to revert a PR from the Head of Design Engineering where he had bumped our Node version up. Someone approved my PR to revert his code, and I merged it. I did this without him knowing because we are in different time zones. However, I anticipated that the next morning, he would want to know why I reverted his code, and I needed to be able to explain myself.

The problem was that I do not have direct access to Cloudflare since I am not on the core engineering team. I didnt fully understand why he made his original PR, or exactly how my deployment broke it. So, I spent a lot of time asking my agent what happened, why it happened, and how I could fix it. I knew that by the next morning, I would probably forget some of the major details, especially since a lot of the troubleshooting happened at 10 p.m.

But I used Entire to help remind me. First, I ran a simple terminal command:

bashentire checkpoint list

This command surfaces a clean snapshot of my history, giving me a list of various checkpoints and commits from my session with the agent.

Normally, I would choose a specific checkpoint from that list and run entire checkpoint explain to get an explanation for each individual moment. But because there were so many checkpoints, I wanted to fast-track the process instead of digging through them one by one.

Fortunately, I have agent skills installed that teach my agent how to interact with Entire commands and execute custom workflows. I handed the agent that same screenshot of my checkpoint list and gave it a simple prompt: Can I get an explanation of each of these checkpoints?

My agent responded by explaining the decisions from beginning to end, covering my initial workaround attempts, how I got confused with the version numbers, and how we finally identified a fix, so the next day I would be able to respond to my coworker.

How to Keep Improving Your Communication Skills

Use agents daily: This one is simple. The more you use agents, the more you will understand how to use them and how to communicate with them. Over time, you will build your own personal workflows that ensure you get the best outcome. It can be a bit of a learning curve at first, but after constant use, you will get a hang of it.
Watch others work: Through reading blog posts, watching livestreams, or even reading my teammates past agent history, I learn a lot about how they operate agents, and I can adapt different parts to my own workflow.
Own the outcome: Never ship code you cannot explain to a human peer. If your agent makes a sweeping change across ten files, trace its steps, audit its intent, and ensure you completely understand the reasoning behind those lines before hitting approve.
Track your agent sessions: Let Entire track your sessions behind the scenes. When you find yourself in a moment where you need durable artifacts because someone is asking what happened and why, you will be fully prepared to communicate it.

It’s a strange world. The best engineers are no longer defined by how well they memorize syntax. At first, the developers who stand out will be the ones who work the fastest, spinning up multiple subagents and impressing everyone with their raw speed.

But very soon, the most valuable engineers will be the ones who can bring clarity back to the humans on their team. When a system inevitably breaks down, they will be the ones who can look back at their sessions, trace exactly what the agents did, and explain the choices to their team.

Code is cheap, but clarity is getting more expensive.

You can get that clarity today with Entire.

Resources:

Website: entire.io
Documentation: docs.entire.io
Discord community: https://discord.gg/jyAENMgmb3

Banning Agent PRs Won't Save Open Source

Rizèl Scarlett — Sat, 23 May 2026 19:52:52 +0000

It's an unspoken rule that large pull requests are poor etiquette. Traditionally, Agile teams break features into manageable slices to make it easier for developers to tackle the logic and for peers to actually review. That is until the adoption of agents. Now developers are shipping end-to-end features in a fraction of the time, but the result is often an unmanageable wall of code that is discouraging to even open. These "Agent-PRs" almost guarantee a reviewer's eyes will glaze over, leading to a quick "LGTM" that misses subtle logical flaws. This tension has become so high that some open source maintainers have moved to outright banning agent-authored contributions.

Before the invention of pull requests, contributors would email maintainers their code changes or ask maintainers to pull updates from the contributor's repository. In 2008, GitHub introduced pull requests, giving teams a structured way to propose, discuss, and review changes before merging them. For almost two decades, the pull request provided the perfect workflow. But now the very structure that enabled collaboration is cracking under the weight of AI-generated velocity.

Prominent open source developers, such as Rémi Verschelde and Jeff Geerling, have taken to social media to express their concerns.

But their sentiment isn’t isolated. Many developers feel this way, and faced with deluge, many have decided to take drastic measures to protect their codebase and sanity by banning AI-assisted contributions completely.

// Detect dark theme var iframe = document.getElementById('tweet-2044406553508274554-404'); if (document.body.className.includes('dark-theme')) { iframe.src = "https://platform.twitter.com/embed/Tweet.html?id=2044406553508274554&theme=dark" }

// Detect dark theme var iframe = document.getElementById('tweet-2011911073834672138-337'); if (document.body.className.includes('dark-theme')) { iframe.src = "https://platform.twitter.com/embed/Tweet.html?id=2011911073834672138&theme=dark" }

// Detect dark theme var iframe = document.getElementById('tweet-2014433315261124760-42'); if (document.body.className.includes('dark-theme')) { iframe.src = "https://platform.twitter.com/embed/Tweet.html?id=2014433315261124760&theme=dark" }

However, as more developers are encouraged to use AI throughout the software development life cycle, running away from AI-assisted pull requests only hurts open source by closing the door to potentially valuable contributors. Angie Jones, my former manager, argues that closing the door isn't the solution. Instead, she advocates for better guidelines for both agents and humans.

She's right. Closing the door isn't the answer, and clearer guidelines are a step forward.

But even with better guidelines, the review process remains fundamentally unchanged. Maintainers are still staring down massive PRs, trying to make sense of what an agent produced.

To remedy this problem, GitHub recently released a feature giving maintainers the power to set contribution limits, starting with a PR cap for outside contributors and an allowlist for trusted ones.

// Detect dark theme var iframe = document.getElementById('tweet-2057864556488306833-376'); if (document.body.className.includes('dark-theme')) { iframe.src = "https://platform.twitter.com/embed/Tweet.html?id=2057864556488306833&theme=dark" }

Many maintainers are excited about this release, but a PR cap is a short-term band-aid. I think it’s a good move, but our industry needs a complete overhaul of how we handle code reviews.

As a maintainer of goose, an open source AI agent, I have spent 2025 and 2026 sifting through pull requests. I do not want to limit who can contribute. I want everyone to be able to contribute. The goal is to understand the context of a PR at a fast pace, even if an external contributor uses an agent to build it. Open source needs infrastructure built to support both human-authored and agent-authored work.

I joined Entire because our convictions aligned: we need to fix the structural breakdown in open source. Right now, we have built a CLI that provides a system of record that captures the context behind agent-assisted code changes. The record gets stored as a This serves as the baseline for a new era of developer tooling that can help our industry move towards:

Shifting from code review to intent review: This means instead of parsing 500 lines of syntax, reviewers will start with intent by examining the prompt, the session transcript, and the reasoning behind key decisions. This allows reviewers to focus on the problem being solved and whether the right calls were made along the way.
The ability to search for the why: Developers and maintainers can ask why a change was made a certain way and receive an answer derived directly from the agent session context.
Infrastructure for the AI-native velocity: The open source community needs a foundation that can support the massive volume of human and agent contributions without falling under the weight of the traffic or causing platform outages.

To explore the full scope of what we are building, you can read about our vision and learn more about where we are headed.

Open source is where innovation happens. It is how large companies thrive, relying on thousands of open source dependencies. Yet if we shut down external contributions out of fear or fatigue, the community disappears, leaving the ecosystem at risk of stagnation. Developers are genuinely excited to contribute because agents have given them the confidence to do so. We need to embrace that momentum by building an infrastructure designed for the AI-native software development lifecycle.

Your Agent Sessions Belong in Your Codebase: Nullius in Verba

Rizèl Scarlett — Tue, 19 May 2026 08:56:46 +0000

Your coding agent sessions belong in your codebase. Before I joined Entire, the company building the infrastructure to bring your agent sessions into your code, I was already exploring this exact idea on my own.

In January 2026, I participated in Genuary, a month-long creative coding challenge where artists, designers, and programmers make and share generative art based on a daily prompt. I used my coding agent, goose, to generate the creative code. For me, this was less of an exercise in creative coding and more of a self-taught lesson in orchestrating agents, since doing complex things with agents was on the rise.

One of the things I built into my process was a repeating workflow where, after every session, my agent automatically committed the session transcript into the same repository that held the creative output. It wasn't elegant, because it was literally a huge transcript with every tool call mixed in and almost no structure to make it readable. I did it because some of the creations were so astonishingly beautiful that I wanted my agent and myself to be able to look back later and have enough context to reuse those same patterns for future challenges.

Three months later, in March, I was working at a company that had built a far more elegant solution to the same problem. Instead of haphazardly dumping whole session transcripts, Entire saves each session as a series of navigable checkpoints. Each checkpoint is a snapshot of a meaningful moment in the session, capturing what the agent did, what changed in your code, and the reasoning that produced the change. Now after using Entire for a few months, I’m realizing that what I had treated as a nice-to-have for myself, I now see as a real necessity for engineers.

I had this epiphany while doing what my job actually entails, which is advocating for developers. I started noticing a pattern across the developers and community members I talked to. While many of them wanted to track their agent sessions, they did not want those sessions to live in the same codebase. Some people felt their sessions were too embarrassing, full of mistakes or moments where they had been harsh with their agent, because all of us have lost patience with a coding agent that just refuses to understand us. Others felt the sessions were too private. Because Entire already supports storing sessions in a separate repository and redacting secrets by default, I assumed we should be louder about that functionality.

Surprisingly, one of my teammates disagreed with me. His philosophy was that agent sessions belong alongside your code by default, and that the discomfort developers would eventually go away. Because I am trained to empathize with developers, I initially felt this stance was dogmatic, and I struggled to see eye to eye. Over the past few weeks, though, the idea kept ringing through my mind, I see his perspective.

Software engineering has never been about flawless first drafts. Our industry thrives precisely because we maintain a transparent, versioned track record of our technical evolution, and when engineers treat interactions with coding agents as ephemeral scratchpads, we end up ignoring a foundational principle of how software actually gets built. Every architectural and logical decision deserves a clear and traceable provenance, and right now that provenance is silently disappearing into chat windows.

I did some historical research on how deeply embedded proof of work is in our industry, and I learned a lot about what happens when we abandon that proof of work. Here’s what I learned.

Proof of Work in Mathematics

This foundation goes as far back as mathematics, the predecessor of computer science and software engineering. In the 1600s, mathematicians operated inside a genuinely toxic environment, settling disputes through public academic duels with brutal stakes. Winners kept their university chairs, while losers were publicly humiliated and often lost their livelihoods entirely.

Because the consequences of losing were so severe, practitioners routinely hid their formulas and hoarded their methodologies. That culture of intense secrecy produced constant intellectual property disputes, redundant reverse-engineering, and a fragmented ecosystem that ended up stalling the progress of the entire discipline.

The turning point came in the 1660s, when the Royal Society of London adopted a new motto, Nullius in verba, which translates to "take nobody's word for it." From that point on, mathematicians had to publish their complete, step-by-step processes in academic journals rather than only presenting final conclusions. In exchange for that transparency, they received institutional validation and undisputed peer credit, and the field finally had a shared ledger of truth.

Proof of Work in Software

Three hundred years later, software engineering experienced a similar reckoning. In the 1960s, code was a tangible, physical artifact. Developers punched holes into cardboard cards, organized them into precise decks, and fed those decks into a mainframe. Version control was physical too, because changing a routine meant pulling a specific card out of the deck and slotting a new one into its place.

Then, code moved to magnetic tape and hard disks and became digitally invisible. Multiple developers modified the same file and accidentally overwrote each other's changes without any shared source of truth. The industry's response was a slow march back toward visibility, moving from local file-locking systems like SCCS and RCS to centralized trackers like CVS and Subversion, and eventually to Git. Git decoupled development pipelines entirely through a distributed, non-linear architecture, but it was hard to use on its own, and it did not pick up real traction until GitHub layered a collaborative interface on top of it. That interface turned version history into a shared social ledger and defined the modern development workflow.

The pattern is the same one the Royal Society set in motion three centuries earlier. Every time our industry has taken a leap forward, that leap has come from making invisible work visible.

Invisible Agent Work

Agentic workflows are becoming the primary engine of software production, but they're abstracting away our work at the same time.

By committing only the final file output of an agent session, we aren't hiding our work the way 17th-century mathematicians did. But the effect is the same: we are back to delivering an end product while erasing the lineage of how it was reasoned into existence.

The prompts you write, the specific files your agent reads, and the back-and-forth debugging it takes to get things right are not just logs. They are first-class development artifacts. When we strip them away from a pull request, the rest of our tooling, our reviewers, and our future selves are all left to take the resulting code at face value.

That is exactly the position the Royal Society found unworkable in the seventeenth century, and there is no good reason to expect it will work for us either.

Nullius in Verba

Including your unedited session next to your code feels vulnerable, but so does pushing your first commit to a public repo or opening your first pull request in an open source project.

That discomfort is not a flaw in the workflow, it is the price of admission for a trustworthy, auditable record of how software actually gets built. Nullius in verba is still the right principle 300+ years later: take nobody's word for it, not even your agent's. Let the work speak in the place where the work actually lives. That is the direction we are building toward at Entire: making the context behind agent-authored work as visible as the code itself.

Did you like this blog post?

Try out Entire: entire.io
Join our Discord: https://discord.gg/WUzRcQ5PX4
Read our docs: docs.entire.io

If you have additional thoughts, feel free to leave a comment!

How to Keep Entire Checkpoints Separate from Your Code

Rizèl Scarlett — Fri, 08 May 2026 08:02:39 +0000

Storing a record of your agent sessions solves the biggest friction point for developers: limited context. On the surface, this may look like a dormant log, but Entire transforms those records into procedural memory. By default, the Entire CLI stores your agent history right alongside your code. More specifically, it stores your checkpoints, snapshots of your prompts, agent transcripts, and the state of your work at each step, on a dedicated branch in the same repository called entire/checkpoints/v1.

But as valuable as that memory is, it raises a valid question: What if I don’t want anyone else to see the conversations I have with my agent?

We’ve heard a few consistent reasons why developers want to keep their agent history private:

The conversations are, frankly, a little embarrassing. (I’ve yelled at my agents before. I am not proud of it, but when tokens are few, so is my patience).
It can start to feel like surveillance from their employer.
It's a privacy concern. Those conversations might include context their company doesn't want to share publicly or with external collaborators.
They want to keep your main repo lean and focused on source code.

If any of those reasons resonate, you have two main paths to a more private workflow.

Push checkpoints to a separate private repo

This is the sweet spot if you’re working on a public or shared project but still want a history that you (and maybe your trusted teammates) can access.

1. Create a private repo for your checkpoints

On GitHub, create an empty private repo with any name you want. In this example, we’ll use myorg/checkpoints-private. This is where all your agent sessions will live. You don't need to add a README or initialize it. Entire will push the first checkpoint branch on its own.

2. Point Entire at the new repo.

From inside your project, run:

entire configure --checkpoint-remote github:myorg/checkpoints-private

The format is provider:owner/repo. Today, github is the supported provider. This writes the setting to .entire/settings.json under strategy_options.checkpoint_remote:

{
  "strategy_options": {
    "checkpoint_remote": {
      "provider": "github",
      "repo": "myorg/checkpoints-private"
    }
  }
}

Now, your code will be stored in your main repo, and your agent sessions will go to your new private repo. You can read more about this in the Checkpoint Remote docs.

Keep your agent sessions local

If you want the highest level of privacy, you can keep your agent sessions local and opt out of pushing them to remote using the following command:

entire configure --skip-push-sessions

This modifies .entire/settings.json with the following values:

{
  "strategy_options": {
    "push_sessions": false
  }
}

This setting still allows you to store your sessions locally. For example, you can still rewind, look back at what happened, and use all the local features. However, because the checkpoints never get pushed to GitHub or any remote provider, you cannot retrieve them if you switch devices. Also, your teammates won’t have access to your checkpoints.

What if I accidentally paste a secret?

We know that mistakes happen, so we have guardrails in place. Whether you store your history alongside your code, in a private repo, or on your local machine, Entire runs every session through a redaction pipeline before it hits git.

We use Betterleaks to automatically scrub:

Cloud credentials (AWS, GCP, Azure)
Source control tokens (GitHub, GitLab, Bitbucket)
Service keys (Stripe, Slack, Discord, Twilio)
Private keys (RSA, SSH, PGP)
Database connection strings with embedded passwords
Bearer tokens, JWTs, and high-entropy strings that look secret-shaped even if they don't match a known pattern

TLDR; Which one should you pick?

The Goal	The Command	Where data lives
Full Visibility	Default	Same repo as your code
Private Collaboration	`--checkpoint-remote`	A separate private repo
Total Isolation	`--skip-push-sessions`	Your local machine only

Note: Redaction runs across all three. Whether your checkpoints live in your code repo, a private repo, or only on your laptop, secrets get scrubbed before they're written.

Conclusion

We recognize that your agent workflow is going to look different based on who you are, the codebase you're in, and your team's unique security needs. Entire is built to adapt to those needs.

Ready to dive deeper into configuring your setup? Check out our documentation on checkpoint remote and Security & Privacy docs.

In the comments section, let me know: Do you even care if people see your agent history, or would you rather keep those transcripts private?

Turning Agent History into Procedural Memory

Rizèl Scarlett — Sun, 03 May 2026 23:40:08 +0000

For about a year, my primary coding agent was goose. Since I worked at Block and served as a Developer Advocate for the project, I was deeply embedded in its ecosystem. I contributed code and provided product feedback that shaped how it functioned.

Then, I moved to a company called Entire that provides the infrastructure for the agentic software development lifecycle. To do my job well, I have to dogfood our product across the agentic ecosystem. This means I am constantly switching between Claude Code, Codex, and other agents that support hooks to contribute to docs, investigate and resolve bugs, understand new features, and produce content.

Switching between AI agents made me realize every agent has tradeoffs. Some are faster or more polished, but I find myself deeply missing a specific goose feature called recipes.

The Problem with Operational Glue

Recipes are reusable, shareable workflows. At the core, they are YAML files that describe a process you want goose to run again. You can write the YAML files manually, but I always preferred the magic of clicking a button to package a successful session into a recipe.

My work in Developer Relations is creative, but it’s built on repeatable systems. For example, writing a blog post, building a code demo, creating a video are creative. The publishing process is not. Publishing a blog post involves a series of tiny, forgettable steps: checking the folder structure, adding the correct front matter, wiring up the metadata, dropping the image in the right asset folder, opening the PR. Each of these steps take a few minutes, but those minutes add up and become hours of operational glue. At Block, I automated as much of that as possible. I had goose generating release notes in CI/CD and creating documentation tickets in Asana. Some of these ran on a schedule, others I triggered manually. The point was always the same: if I found myself explaining a process to an agent more than once, it was an operational smell that needed to become a reusable asset.

While my use cases focus on content and community, this pattern is universal. In many fields, employees find themselves frequently explaining the same sequence to an agent, so why not automate this into repeatable workflows?

For engineers, those repeated conversations may look like:

Upgrading a dependency safely
Bootstrapping a new microservice
Triaging a production error
Writing a design doc or RFC
Preparing a release PR

The inputs and thinking may vary, but the process: conventions, the file paths, the validation steps, the commands you run, the people you always notify stay the same. And this level of automation is necessary today where employers are demanding more output.

The System of Record

I'm constantly jumping between different agents. Each one has its own process for automating workflows, but none of the automation tools hit the mark for me like goose did. While I don’t have access to my treasured repeatable workflows anymore, I do have access to the unique and valuable agent session data that Entire collects. Entire is a CLI-first system of record for agent-assisted development. It captures the context behind your work: the sessions, prompts, responses, tool calls, file changes, and Checkpoints. A Checkpoint is a specific moment where work is tied back to git. It connects the "why" of the agent session to the "what" of the final commit.

I realized this data isn't just for review, audit, or to sit quietly in the background. It's a source of truth that can be used for building better workflows. I thought, “What if I could use my Entire session history to recreate that ‘package up a session’ magic, but in a way that works across any agent, and works retrospectively?”

The most popular way people are currently building reusable workflows is with Skills, so I built an orchestrator skill called Session-to-Skill. It creates Skills for me based on repeated behavior.

The Before and After

Before I used to say:

“Look at past blog posts in this repo, check the folder structure, and the front matter.”
“I want to add a new blog post. Here’s the content: [insert content copied from google doc here] ”
“Create a new PR. Make sure we’ve pulled the latest from main and branch off main before you create this PR.”
“Why did you make the word Checkpoints lowercase when I purposely had them capitalized? Please restore that.”
“Does the OG image work? What’s the path for me to check that again?”

Now, I can say:

“Create a blog post from this content [insert content copied from google doc here].”

This is possible because I prompted my agent to use the Session-to-Skill Skill: "Look at my past sessions where I set up blog posts. Find the repeated steps and conventions, then draft a Skill from that data, so I can create blog posts quickly in the future." My agent created a Skill called Create-blog, which included requirements to properly format the blog, open a PR, and return the path to confirm the OG image rendered.

Well, that’s kind of dumb..

Some may have pushback on this idea of me building an orchestrator Skill because at any moment in a session you can prompt any agent to turn it into a Skill.

The reality is I don’t have perfect foresight. Most reusable workflows are recognized later. After the third time I publish a blog post, I realize I have been doing the same thing over and over again. By then, the valuable evidence is spread across past sessions.

There is also the issue of quality. Asking an agent to summarize a transcript often leads to overfitting and noise. The resulting Skill might include accidental details, temporary file paths, or one-off preferences that happened to be present in that single session.

Instead my Skill is extracting the answers to the following questions:

What was the reusable behavior?
What should a future agent know before attempting this again?

I don't have to remember the session ID from six weeks ago. I just know the work happened. The Skill uses Entire to search my session metadata, checkpoints, and explanations of prior work to find the durable pattern.

Procedural Memory as Infrastructure

My approach creates procedural memory for agents. Procedural memory is the answer to the question, "How do I do this kind of work well, here, in this repo, with this team?"

Daily engineering work is not net-new. You may receive a new ticket, but somebody has solved this problem before.

By using Entire's data to generate Skills, I get a layer of determinism and portability. The agent starts with a template based on real work rather than a generic prompt. It encodes patterns that have already succeeded. And because Skills are portable files, I can take my blog-publishing Skill from Claude Code to Codex without re-explaining my workflow and share it with teammates.

With all this said, I want to urge readers to stop treating our agent sessions as disposable and start turning our history into our infrastructure.

Check out Entire at entire.io

I Don’t Make Slides Anymore. My Agent and Entire Do It for Me.

Rizèl Scarlett — Sat, 25 Apr 2026 17:00:18 +0000

Signing up to speak at conferences is fun until the conference date starts approaching and you realize you still have to write and practice your talk. For me, writing the talk isn't the hard part. I have a process of talking to myself on a peaceful walk (or even in the shower), recording my voice, and then inserting the demos afterward. The part I often procrastinate is making the slides. Creating slides used to be fun, but as I’ve grown my career and my family, it's no longer a good use of my time.

I've looked for a way to automate slide generation, but most options have been fragile. They generally struggle with formatting and taste. However, a few months ago, some of my teammates at Block discovered the frontend slides skill and introduced me to it. This skill enables agents to build out HTML presentation decks that can be exported as PDFs or PowerPoint presentations.

Agent Skills

If you're not familiar, agent skills are markdown files that provide instructions for the agent to understand how to use a tool (i.e., a CLI or an MCP server). This way, your agent immediately knows what commands to run and how to navigate the tooling when you make a request.

How I Use the Frontend Slides Skill

As I mentioned, I already have my talk transcribed, which gives me a talk track to follow. I typically give the transcript to an agent and ask if there are any parts that don't make sense or any gaps for the audience.

Once the talk track is polished, I give the copy to my agent and prompt it to use the frontend slides skill to build a deck based on the track. I prefer to use Claude Code for this task, as it seems to work really well with the frontend slides skill, but any agent that supports skills should work. The agent then produces a beautiful slide deck for me. It really doesn't look bad or overly generic at all. It has various themes to choose from, and since it's generated with HTML, CSS, and JavaScript, I can prompt my agent to edit parts like making the font bigger, changing colors, and so on.

Here's an example:

// Detect dark theme var iframe = document.getElementById('tweet-2024060044153692197-26'); if (document.body.className.includes('dark-theme')) { iframe.src = "https://platform.twitter.com/embed/Tweet.html?id=2024060044153692197&theme=dark" }

My favorite thing to add is a presenter view. It doesn't generate that view by default, but I do like to take a peek at my notes as I speak. So I usually tell my agent to implement that view if I press a key like the letter "P," and I make sure it syncs with what everyone else can see. Then, I upload my deck to GitHub Pages. Goodbye, Canva, PowerPoint, and Google Slides.

How Entire Enhanced My Workflow

Let me rewind for a second and introduce you to Entire. Entire is the company I work for. We're building the next developer platform for the AI-native software development lifecycle. The team recognized that agents have changed our workflows, so the infrastructure we use should change too.

Our first tool is a CLI that captures prompts, agent responses, tool calls, and other session data from the work you do with an agent. That gives you a way to inspect what happened, rewind work from a past session, and stay accountable. For example, if a production outage ever happens, instead of saying, "Oh, the root cause is that my agent did it," you can actually track the decisions made between the agent and the person prompting it. I described this to someone at a conference the other day, and they boiled it down to version control for agentic work, which is honestly exactly what it feels like.

Now, it took me a while to see how Entire could make my workflow even better, but the founders opened my eyes. I can use Entire while I'm building out a demo and then use that captured work to help build the slide deck later. Entire has a command called entire dispatch. It generates a markdown summary of the work Entire captured between you and your agent in a repository.

For example, after experimenting with OCR in a repo, Entire generated this dispatch summary for me:

# Dispatch: blackgirlbytes/pretext-handwriting-demo

Shipped a full handwriting recognition demo built on Pretext, moving from  
initial scaffolding to a polished scrapbook composition surface within a  
single day.

## blackgirlbytes/pretext-handwriting-demo

### Handwriting Recognition

• Built draw-mode handwriting recognition as the core interaction surface.  
• Added image upload as a second recognition path alongside the drawing  
canvas.  
• Implemented auto-recognition after drawing completes, removing the manual  
trigger step.  
• Added camera mode to round out the three input methods.  
• Removed the explicit draw button to streamline the input UX.

### Scrapbook Composition Surface

• Introduced a scrapbook composition surface for arranging recognized text  
and shapes.  
• Integrated Pretext to handle obstacle-aware text flow around placed shapes.  
• Added animated motion layer to the scrapbook background.  
• Added resize handles to scrapbook shapes for direct manipulation.  
• Matched scrapbook background typography to the handwriter component for  
visual consistency.  
• Moved scrapbook controls into the composition header to consolidate the  
toolbar.  
• Fixed scrapbook layout and output tab rendering, then simplified and  
tightened tab spacing across multiple passes.  
• Corrected background line wrapping and ensured background renders before  
shapes are placed.

### API Key & Session Management

• Added session-based OpenAI key setup to avoid requiring environment-level  
configuration.  
• Hardened API key gate interactions to handle edge cases more reliably.  
• Added environment key setup path as an alternative to session entry.

### Documentation

• Added project agent working rules and intent guidance early in the commit  
sequence.  
• Documented project setup and architecture in the README.  
• Refined and clarified the README project description across two follow-up  
commits.

All core features landed on main on April 1, 2026; the repo is in a  
reviewable state.

That's helpful for me because building a demo usually takes a few days. I'll leave it, come back to it later, and then have to remember what I actually did, what mattered most, and which technical details are worth calling out. Instead of trying to reconstruct all of that from memory, I can use the dispatch summary, give it back to the agent, and ask it to make a strong slide that captures the main technical highlights of the demo. That saves me from having to recall every step I took days or even weeks later.

Beyond DevRel

My particular use case works best for folks in Developer Relations or folks who do DevRel-related work like conference speaking, but this can actually work well for various roles. Here are a few:

Developers demoing completed features to their team. My husband is a developer, and he's expressed that building the slide deck is time-consuming just to show off a feature he built to his team.
Hackathon participants demoing their project to judges. Presentation decks often get neglected because everyone is focused on building the actual project during a small window of time.
Solutions engineers or sales engineers preparing customer demos. A lot of time goes into building out the demo environment itself, so having help turning that work into a clear deck can save a lot of time.
Workshop instructors or developer educators teaching technical material. It can be useful to turn the work captured while building the demo or sample app into slides that explain the flow, architecture, or key takeaways.
Engineering managers or tech leads giving project updates. Sometimes the hard part is not the work itself, but summarizing what happened clearly enough for leadership or cross-functional teams.
Founders or indie hackers pitching what they built. When you are moving quickly, the last thing you want is to spend hours making slides after already doing the hard part of building the product.

I don’t believe in automating things that deserve a human touch, but I do believe in automating things so I can spend more time with humans. The slides skill has been great for me, but adding Entire to this workflow has made it even easier for me to do that.

Much of my previous work with GitHub and Block (goose) was focused on using agents to build faster. Recently joining the team at Entire has pushed me to think more about the next layer: making agentic work durable and accountable.

Building fast with agents is fun, but in practice, I also need to be able to understand what happened, pick work back up later, explain it to other people, and turn it into something useful beyond the moment it was created. I’ll be sharing more of my agent-native workflows as I continue experimenting.

If you want to learn more about Entire, check out our:

And follow me!