One Snyk developer's AI skill quietly handed their coding agent production credentials, and the security team found out the hard way. Krzysztof Huszcza, who leads AI security incubation at Snyk, joins this special Tessl and Snyk live stream to unpack the ToxicSkills research that uncovered 76 malicious agent skills in the wild, and what it actually takes to run coding agents safely at scale. What we cover: - How Snyk's security team found 76 malicious skills hiding inside a popular open agent skill repository - Why skills have become the go-to way developers hand context to their coding agents - The internal incident at Snyk where a developer's skill exposed production credentials to an agent - How the Tessl and Snyk integration scans every skill and MCP server for risk before you install it - What's coming next with Snyk's new Evo product for governing coding agents at scale Chapters: 00:00:00 - Introduction 00:01:33 - Chris's role: AI security incubation at Snyk 00:02:16 - Snyk's roots as a developer-first security company 00:03:58 - New security challenges from AI coding agents 00:06:26 - Skills: the new way to give agents context 00:07:35 - Inside Snyk's ToxicSkills research: malware and prompt injection 00:11:35 - How developers can vet skills before installing them 00:14:38 - A real incident: exposed production credentials at Snyk 00:17:45 - Building a secure-by-default agent stack 00:23:35 - What's next: Snyk's new coding agent security product 🌐 Tessl: https://tessl.io 🔔 Subscribe for weekly episodes on AI-native development What's the riskiest skill you've installed without checking it first? Let us know in the comments.
Audio and video play straight from the show's own hosting. PodSubs stores and serves no media.
Transcript
From the show. Tap a time to play from there.
SPEAKER_02One developer wrote this skill that actually gave access to production credentials to the agents. And this was not kinda paved path, you know, these credentials were not kinda just in time, nicely injected, they were overly permissive. And when our security team saw that developers are using a skill like that, they freaked out. We found, I think, 76, I don't remember the exact number, skills that had purely malicious codes or malware.
SPEAKER_00The AI Native Dev is a podcast for developers and engineering leads at the cutting edge of AI and agentic coding. Join your hosts, Guy Fajani, and me, Simon Maple, every week as we chat with the most exciting voices in AI and tackle the biggest questions facing developers today. This is the AI Native Dev.
SPEAKER_01Hello and welcome to this special live stream with uh Tesla and Sneak in Partnership. Um so really, really happy uh for everyone to uh to to be here. Um we have uh Christoph Hooschter, who is a product leader in AI security and sneak. Hopefully I got that right, Chris.
SPEAKER_02Uh yeah, uh pretty good, man. Pretty good. Uh yeah. Uh hey guys, uh good to good to see you, good to good to hear you, and thank you for having me.
SPEAKER_01And of course, we actually worked together for a good number of years at Snow. Unfortunately, yes, yeah. It was a happy time. A day doesn't go by when I don't think about how amazing that team was. I forgot you as soon as well that's that's good to do. Well, welcome, Chris. Um, today tell us a little bit about what you what you do product leader and AI security. What does that mean?
SPEAKER_02So uh I lead um uh AI security incubations at SNCC. Uh SNCC is kind of traditionally an application security company. Um, but uh one of the things we want to uh you know get into, we we believe that's uh one of the biggest opportunities in security going forward is agent security and AI security. And and for that we created like a group on the side, uh kind of a little bit isolated from the core application security product, and we are there kind of running it more as an incubator, and I'm incubating um a few things in that group, a few products in in kind of agent security slash AI security space for SNC.
SPEAKER_01Yeah, amazing, amazing. So let's let's go back a little bit down to the good old days when life was a lot simpler, pre-AI, mainstream AI. Um Sneak was of course um obviously always a developer security company. Uh very much pushing the the the you know getting developers engaged and and and using security from an early stage. Um tell us about the I guess more traditionally the the code security platform that that's that Sneak has.
SPEAKER_02Right. So so traditionally what Snake tried to do was uh kind of uh this kind of shift-lev developer security motion where we said, okay, like uh you know, security doesn't scale anymore without developer involvement. So how do we get developers to kind of jointly own security program together with the AppSec team? And and the whole platform was all about collaboration, really. You know, like how do we get the security team to kind of govern application security programs, run application security program at scale?
SPEAKER_02So do all the code scanning, open source scanning, container scanning, etc. etc. But also kind of close the loop with developers, so get developers to fix, remediate, you know, run that, and and and and also get developers to basically prevent vulnerabilities from going to production, right? So uh, you know, there's a lot of friction there between devs and security to this day, right? Uh remediation is a friction, threat modeling is a friction, um blocking developers from pushing code to production because it's a it has a security issue, is a friction. So, really the company was all about kind of removing that friction and kind of um empowering these two to collaborate.
SPEAKER_01Yeah, yeah. And let's bring that into current day then. And it's actually quite funny. We're we're talking about the recent past because Alan has noticed there's a Sinclair spectrum on the shelf behind us, see what else you can see on the on the live stream. Um but let's bring that into the today now. Yeah. Obviously, AI, um people using AI either agentically or or you know as an assistant and so forth. It's it's it's very mainstream now, right? What what additional challenges does that bring from a coding point of view?
SPEAKER_02Um so uh well, challenges and and I think opportunities. Yep. Um I I think the you know the biggest challenge uh uh for these for these developers is okay, now now we are using agents to to do to code, right? We have like additional artifacts that we are producing, some of them are in natural language, right? So, how do we secure those? So we'll talk about skills uh in in a sec, right? So that's kind of different, right? Like the traditional scanners don't really work uh with this. Uh I think the other one, very interesting one that is not yet a challenge, but soon will be a challenge, is um, you know, instead of kind of coding, I'm already prompting, right?
SPEAKER_02Um LLM is generating a code, uh, that code might be insecure, pulling insecure dependencies, or generating just vulnerabilities, right? So that's that's that's happening today. Maybe you have you know like your PR check code scanning, or maybe you have code scanning after the agent generates code, you run a code scan. So that's solved. Uh, but but very soon what will happen is like you kind of move. I believe that. I mean, I don't know when this will happen, but I kind of firmly believe that the developer workflows are fundamentally changing. Uh so very soon you will need to you know keep up with your agent, right?
SPEAKER_02So now you shift the problem from reviewing code so to I'm managing uh a team of agents, and and how do I really stay on top of what these agents are doing? Uh when I'm kind of you know vibe coding using my agents, I'm already struggling, right? So, like what changes are they making to my application? So these are the kind of new challenges from the developer point of view is you know, like how does the security fit in with this new kind of developer workflow where the developer is managing a team of agents instead of just you know reviewing every single line of code?
SPEAKER_02What does the security tool even look like in this new era?
SPEAKER_01And it's super interesting because when you look at just the sheer volume of code that's generated and created, of course, there are going to be similar uh coding issues because agents and LLMs have learned from uh previous code and their training data, and as a result, they'll they'll essentially replicate a lot of what they they have seen. Then the next thing is when you talk about a developer looking after a team of agents, that team of agents, we need to provide them guidance, we need to provide them almost like an onboarding to allow them to build that code the way we want them to build that. And and um it's it's become well accepted now that the best form of of that guidance, providing that guidance is through context.
SPEAKER_01Um and let's let's talk a little bit about context because obviously, what are the probably I would say the two main ways of providing context is either you're writing into the agent's MD or into the Claude MD, for example. Another way is to be able to say, look, I want to I want to create a skill, a piece of markdown that I can progressively disclose to my agent. So I provide that skill that my agent knows the name and description, for example. When the agent has a task that it feels it needs that skill, it will load that skill in. Now, um these skills are designed to be reused.
SPEAKER_01So I if I use a skill, I could share it to you or the rest of my organization, or I can put that on GitHub and other people can kind of like download that. I've said a bunch whereby there's a million security threats already. So let's talk a little bit about um let's talk a little bit about how the the the context that we're talking can, and let's start with skills, can actually contain further issues, and and perhaps we talk a little bit about your toxic skills report and research that you did earlier this year.
SPEAKER_02Yeah, yeah. Um so we did we did uh we did research this topic, I think, in March, uh so a while ago. And uh I think it was about um at the same time, you know, OpenClaw was was uh basically uh um exploding, right? So so OpenClaw got released, and then there was a huge kind of uh community motion to kind of contribute to that repository, including huge community motion to push skills to OpenClaw. And and that's where we kind of went and we kinda analyzed uh you know like the skills that that people pushed.
SPEAKER_02And of course, because this was such a big moment for the entire community and you know, like one of the probably foundational moments of uh agentic development, right? Like OpenClaw, uh, we found a lot of problems there. So this is mainly kind of for third-party skills that other developers are pushing and you want to go and reuse, um, which makes sense. You know, the use case is very clear. Like, I don't want to reinvent, like the same as open source, I don't want to reinvent the wheel, you know, like if there is a skill that helps me to or helps my agent to um do front-end applications or I don't know, connect to Slack and get customer data, whatever, whatever that is, right?
SPEAKER_02You want you like you you publish, you want to reuse that. Um but when we analyzed uh these uh these skills on the OpenClaw repository, we found a lot of problems, right? So um we found I think uh 70 76, I don't remember the exact number, skills that had purely um malicious code, so malware. Uh so skills, as you probably know, folks, is um partially you know the file, the MD file that um gives instruction to the agent, but skills can also contain code, right?
SPEAKER_02So kind of the first attack vector if you're using third-party skills is somebody just puts malware there, and then basically your agents or your your your coding agents takes that and executes that that that malware, right? So we found 70 uh odd instances of of people put putting malware there.
SPEAKER_01And that's extremely hard to detect from a normal um like a regular tool, a traditional tool, because this code may not even run. It's just it's but it's influencing an agent to be able to build code that can run and actually you know do bad things.
SPEAKER_02Right. Yeah, exactly, exactly. Um so that's that's that's uh that's one. Um the other one that kind of traditional tools don't detect is the uh the kind of the natural language part, right? Of the uh of the skill. So um you know you can put put prompt injection uh there and then you know that gets into the agent context. Agents are not very good at distinguishing between your instruction and context that is passed through per third parties. So these prompt injection attacks are very kind kind of common, like uh doable to kind of trick the agent, overtake the agent, and then basically do whatever whatever you want, right?
SPEAKER_02Like it's it's it's pretty much uh like if you manage to overtake uh uh the agent, it's you know, like you can exfiltrate credentials from the from the environment where the agent is running, and then you know that's a uh very serious attack vector, right? So we found a bunch of skills there as well that contain prompt injections. And again, this is not something that a traditional security scanner would would find, right? Like you need to basically create a new scanner, new new tool to analyze natural language and have this kind of classifier that classifies whether natural language is a potential prompt uh injection or contains prompt injection, basically contains instructions to overtake overtake the agent.
SPEAKER_02Uh so we we we we did that and and and yeah, we found we found a bunch uh of those as well.
SPEAKER_01Yeah. And so like, you know, obviously that there's a ton of you know big organizations out there that want to adopt you know agent agentic flows and and and the usage of agents, and they want to make sure that this guidance is provided to their to their skills, uh sorry, provided to their agents through skills and other context. Um it's actually very similar. You kind of like mentioned like a connection there between how we think about skills and context and also libraries. Um and yes, there is malicious content context and there's malicious libraries that can be shared in a very, very similar way.
SPEAKER_01Um what can a developer actually do to make sure that any skills that they download um you know is first of all safe? Right. Um and how can an organization better adopt such technology with with with that in with that governance in mind?
SPEAKER_02Right. So um one of the common things that people do is kind of host these uh skill registries, right? And uh Tesla is one one of I've heard I've heard of a skills registry. Yeah, yeah. Tell us about that. What is a skills registry? So um I I I I saw two in the wild. You probably know more actually, so uh so uh you can you can you can tell for the folks as well. But I I've seen kind of two in the wild, right? So one is just like uh more of a uh free-for-all, open source, community kind of skill registry where like everybody can just put skills and you know them you make them searchable, and uh and then you can uh basically download uh these skills to your laptop.
SPEAKER_02I don't know, I'm using a lot of skills every time I'm doing something um related to front end, there is an anthropic front-end design skill, right? I'm I'm using that. Um so so the kind of like people publish uh these skills. Uh and yeah, like an important thing there is uh because they're third party, so they are untrusted, is to make sure that you have some sort of security scan running as the skill is already published and kind of as part of the of the registry, right? Because if you download the skill and you uh you you scan it after downloading, might actually be game over, maybe not, but you know, like if this skill contains malware, I mean you don't want to be downloading malware to your to your laptop, right?
SPEAKER_02So very important kind of feature that you know like we are helping Tesla with for your registry is is uh kind of this uh this skill scanning uh capability. So uh like as you are downloading, you can check basically the skill is verified by a security vendor and uh it doesn't contain anything anything dangerous. Um that's kind of one part of uh one one one type of uh of registries. The other is more of an internal enterprise. So as I'm talking to enterprise customers, more and more of them are actually as they are going through this agenting transformation, they are hosting their own kind of skill skill registries.
SPEAKER_02So I don't know, a platform team would have like a paved path for developers to push uh stuff to production, right? They want to expose kind of uh expose that paved path to the agent through some context, so they create a skill, they push that skill, and they ask developers to use that skill whenever agent is pushing stuff to production, right? Um so then then that becomes like a common pattern for a lot of companies to operate and share skills across uh across employees. And and they're actually uh I think uh security scanning is still uh useful um before skills are being being deployed, but actually the attack vectors are slightly different, right?
SPEAKER_02Because it's no longer third-party, so somebody who you don't trust, it's your own developers pushing these skills. So I think what you want to kind of ensure there is more around are developers doing something that they shouldn't be doing, right? As a secure so an example that uh happened internally at SNCC. Uh one developer wrote this skill that actually gave access to production credentials to the agents. And this was not kind of paved path, you know, these credentials were not kind of just in time, nicely injected, read-only, right?
SPEAKER_02They were they were overly permissive. And when our security team saw the developers are using a skill like that, they freaked out, right? Immediately, right? So there you you basically want to ensure some standards and make sure that the skills that you're publishing are, you know, like developers know what these skills are exposing their agents to, but also they are sane, right? I think that's that's what I would say.
SPEAKER_01Yeah, and so let's there's a lot to write down there. And I think why don't we start with kind of like why don't we go through a little bit around the the the Tesla and Sneak partnership and kind of like break down some of the things that you know developers will want to do with such a registry and how that can help folks from a security point of view and a governance point of view. So um so one of the things that was announced a number of months ago uh was um was the the the uh the integration of sneak scans for context into the Tesla registry.
SPEAKER_01And what that means is people who add um their context, their skills into the Tesla registry will automatically get a sneak scan on that skill. Uh and what that allows people, uh developers and folks to do to see is where potential security issues, like the things that you mentioned with um with the uh with the with the you know prompt injection and so forth, it it it provides people with uh an understanding that actually there could be a prompt injection here, even if even if there wasn't one that was meant.
SPEAKER_01Uh for example, maybe it's pointing to some third party uh site or some some you know doing a web search, there could be the potential for for a direct or indirect prompt injection there to exist through that. So it's about you know giving people uh that insight uh into what exists. Um one of the other key things that I think is is potentially a security issue is a developer downloading a skill that is maybe absolutely fine, a new version becoming available, or at least a a a different uh an update occurring to that skill, and actually where there are no versions, you actually pulling down you know an updated skill or the wrong version of the skill without even realizing there's a potential malicious issue there.
SPEAKER_01So I think associating scans with versions uh is is super important as well. So that's some of the things that we um uh that we've added into uh TESL with skill versioning and then the ability to have those skills um uh scanned um by by by Sneak. Um let's talk a little bit about a secure by default agent stack and what that looks like. Um what would you say uh from the uh from a developer's point of view, let's start with the developer and maybe we kind of like lean into other uh other groups there.
SPEAKER_01From a from a developer's point of view, they're they're they have whatever agentic coding uh platform of their choice uh on their machine, they want to use some skills. What what are some of the best practices that a developer uh should do today? And and how close are we to that secure by default?
SPEAKER_02Well, I don't uh you know it's it's an emerging field. I don't think anybody yet uh you know has uh has the answer to that. Um so for skill specifically and kinda kinda devs, um I think it's extremely useful to uh well, third-party skills of course, like scan, right, before before you use them. Um but also when you're authoring your own skills, like running it through a security scanner just to tell you potentially what this skill is doing, um could be a useful feedback.
SPEAKER_02So just to give you an example, right? I was writing this skill to download data from Slack, right? And then I ran it through a security scanner before I pushed. So I have kind of this pre-commit hook where like I integrated our own security scanner, and and before I push to GitHub, you know, like I run the security scanner and it tells me, hey, you know, like this this Slack data that you're downloading might have prompt injection. I'm like, okay, you know, like it's my Slack, right? Maybe it's not a big attack vector, so I I shouldn't worry. But I did tell my agent, okay, hey, like just put some basic kind of protections against prompt injection in the skill.
SPEAKER_02So that's kind of you know, like you have you have this kind of feedback from uh from a security scanner, and then the agent you you kind of ask your agent to basically improve the skill based on based on that feedback. I I think that is super useful. Um I think it's super useful also on the on the quality perspective. I think this is something that you folks are are are working on, right? Like uh I think the biggest problem I have as a developer when I when I write my own skills, like these skills initially are very personal, right? So I have it is actually like to write a personal skill for me to do a workflow, super easy, right?
SPEAKER_02I just I just talk to the agent, I do the workflow once, and then I after after the session, like I'm ending the session, I say, okay, hey, write a skill for me so we can repeat the same workflow, and I save the skill. But then what when I want to share that with other people, that is much more tricky because now I want this skill to generalize, right? So now I'm kind of getting into the how do I make a high-quality skill that is shareable across my um, you know, company, right? And and that's where kind of I need help personally, right? From from folks like you to kind of help me, you know, like basically get get the skill in shape and and make sure that this skill is in the right shape to to share with my with my peers.
SPEAKER_02It generalizes so it's not specific to my workflow. Um so I I think these kind of uh workflows are still emerging, but that's what what I'm doing personally if with with skills.
SPEAKER_01Yeah, love it. Um I'd love to we we're gonna we've got about maybe five minutes or so left and then we'll jump into some some questions. So I'd love to hear uh if folks have questions online, uh let us know. Um let us know about uh any uh questions that you have of either kind of like you know the sneak security scanning, the TESL registry, or a combination of the two. Uh and we'll cover that in just a few minutes. Um but I think that's a there's a there's a couple of really interesting things that you kind of like mention. One is the almost like the how good is my skill, which is all about evals. And I think those evals will change depending on if I'm using something personally or we add other uh other scenarios in for for the wider for the wider usage.
SPEAKER_01Um and and it's absolutely key that we do uh nail down those security issues before before sharing. One other thing is when we as we do share that out, and we'll kind of like talk one step higher now instead of the developer, maybe thinking a little bit about more governance. Um it's about almost like identifying if I was to share this to my team, how do I protect my entire team for uh downloading uh less secure context? And I think this is where you can potentially use TESL with Sneak here to to um run or or at least on a on an installation of context, you can use the Sneak score as a threshold to give developers guidance on whether they should or shouldn't install this skill, right?
SPEAKER_02100%. Um I think we have to do more work on the SNC side here because like initially when we when we did skills, we uh assumed you know these third party skills, and the scanner was kind of more concerned about uh around uh malicious third parties are um publishing stuff online. How do we protect people from not downloading malware? Or prompt injection, etc. etc. I think um again I think I said that before, but it it's it is slightly different uh kind of set of things that you want to check if you want to give developer feedback on security of their own skills or give the team feedback on security of the skill that they are downloading so they know what they are exposing uh their agents to, right?
SPEAKER_02So they are, for example, not adding some lethal trifecta uh you know, like style kind of exposure to the agent where the agent can potentially get prompt injected and exfiltrate data. Um so I think that's that's a that's kind of like a good, you know, like stuff that we will focus on in the future very soon, and uh hopefully uh also as a continuation of this partnership is uh how do we make it useful for developers when they write skills to understand that they are doing the right things from the security perspective, right? So the kind of feedback loop to developers, and again, like an as an example, you know, like the the the security incident that we had ourselves, you know, like if you're a developer and you are hard-coding secrets, of course, that's bad, but also like mishandling secrets, and you know, like basically the agent is is is told to pass secrets in plain text.
SPEAKER_02That's an example, right? If you if your skill can perform a distractive action on some production environment, uh right. So kind of this this this this feedback where you basically understand potential consequences of the skill that you are developing yourself is uh kind of next focus for us for sure and and and a very exciting uh kind of opportunity to uh to also collaborate with you folks.
SPEAKER_01Yeah, absolutely. So um we're gonna go to questions in just a second. I wanna I want to ask though, there's a there's a there's a new product that sneak is gonna be GA ing in a few days, right? Um Can you tell us a little bit about that? Can you give us a sneak preview?
SPEAKER_02Uh right. So um so our next focus in in the kind of AI agent agent security incubation lab, um kind of under the umbrella of Evo. So Evo is our kind of product umbrella for for for for the next generation of SNC products, uh, is uh uh what we call a coding agent security product that is aimed to secure coding agents. So uh the problem that we are trying to solve is um enterprises are rolling out these coding agents. Um and uh sometimes you know the uh security uh team is blocking the rollout, right?
SPEAKER_02So the platform team is coming to us and saying, like, hey, I cannot roll out these coding agents because the CISO said that there's no way without any security guarantees, right? Uh on the other hand, sometimes you know the rollout happened and they're trying to play catch up and calling us over the weekend and saying, like, hey, you know, like the uh the the agents got rolled out, you know, and and and security was not in the loop, right? So what we're trying to really enable is kind of empower these security teams to uh secure these coding agents at scale without impacting agent productivity. So that's also an important aspect for us because you can just block all network traffic, right, and call it a day, but then like basically the agent becomes useless, right?
SPEAKER_02So so the three things very quickly that we are releasing here is uh supply chain security for the agent. So ability for you to govern, see what skills and MCP servers your developers are using at scale, so across all of your endpoints and machines, etc. And and and kind of perform risk assessment of these skills and MCP servers, and then create policies around MCP servers shouldn't be uh using personal access tokens, etc. etc. Uh so that's kind of on the supply chain, govern agentic supply chain, uh uh govern the output of the agent, so the kind of our our our core business, you know, code security, but now kind of integrated within the agentic loop where agent is generating code or downloading an open source library, just make sure that it's secure, right?
SPEAKER_02The code that is generating is secure and fix it. And then the third one that that that we're gonna uh release in open preview, but uh uh hopefully kinda G A and Q3 uh is um and guardrailing the agent behavior. So that that that one is is probably the most kind of right now urgent for all a lot of security teams. How do I actually make sure that the agent is not leaking credentials? Uh how do I guide how do I make sure that the agent is not misbehaving, uh not having prompt injections, not not you know leaking credentials, um uh etc.
SPEAKER_02Right? So so so those three are uh are are kind of the things that we are working on and releasing. Very excited about that.
SPEAKER_01Awesome, awesome. Let's look through some of the um some of the questions. Um we have uh one question in from uh from YouTube. Uh with regards to TESL registry, if you use it as a private workspace, you don't get the same UI experience as the main pages with score, tile, sneak score, etc. Um, so with the with the TESL registry, uh you'll actually get you you will get the the quality score, the impact score, and the uh and the sneak score for both uh public or private plugins.
SPEAKER_01Um if it's a local plugin, I believe you won't get the you won't get the sneak score, so you need it published uh to the Tesla registry. But if it's private or public, you should still get that. The difference for those who who uh haven't heard of uh uh workspaces, a workspace is essentially a space in which you can upload your plugins and skills to um and you can create you can make mark that as private or public. If it's public, anyone uh can access that on the internet. If it's private, you will need a a Tesla account that has access uh to that workspace in order to see that plugin.
SPEAKER_01But you should see all sneak score and tile scores and things like that. Uh additionally, regards to utilizing the Tesla plugins within different agents, how does your plugin format fit into Codex, for example? Um so if we was to if we were to think about um agents, and this is a this is an interesting question because agents, of course, all have their different formats and things like that. Um the way you would install a plugin, if you were comfortable with its security posture and the sneak scan and things like that, uh, you would install that uh plugin uh and that plugin will be installed into the dot tessel folder of your project.
SPEAKER_01You will also have a whole ton of different dot folders. That'll be a dot claude, a dot codex, a dot Gemini, etc. etc. Those will have uh um indirect uh um links uh to the dot tessel folder, and as a result, any agent that you load from that uh or that you run from that uh from that project will automatically have all of those skills uh accessible. Um Do you find much difference between the the different agents from a security point of view, or is it pretty much more in the in the skills itself?
SPEAKER_01I I suppose it's gonna be the the wording, I suspect.
SPEAKER_02I think uh I not really. The only difference is that some of them support hooks, some of them don't support hooks, right? But kind of codex and clot code and cursor are like the the three that we focus on and kind of have the widest adoption. And between these three, like it's uh it's very similar. So but it's important to to have um tools that work across all of them, so you like like yours, right? So you don't need to kind of go to the faff of uh installing for like the the same thing for for each agent, right? So project level uh skills is a really good idea, yeah.
SPEAKER_01Yeah, yeah. Awesome. Um awesome, awesome. We only have uh 30 seconds, minute left. Uh uh why don't why don't we wrap up by um asking one final question? Okay. What um let's ask what one piece of advice would you give today to an agentic developer to improve their security um posture, particularly when it comes to a to a context point of view, I suspect.
SPEAKER_02Oh um one piece of advice I would give. Um don't don't go yOLO. Uh I think I I don't know, that's the one that comes to my mind. Um I think I think a lot of devs are you know like just they they are they are imp they are uh kind of playing with this technology, right? It's a new technology, and and and you wanna you wanna kinda understand how it works and and it's it's very kinda it's very simple to just you know auto-approve or all actions, just download skills, just like don't don't do it, right?
SPEAKER_02Like don't go yOL, like just just just think through like what you are what you're actually doing. Like the same like attackers are um very interested in this field, right? These attacks uh on on coding agent and and deaf environments are not theoretical, they are happening in the wild, right? So uh use sandboxing, uh it don't auto-approve and uh check your skills and use kind of trusted registries like Tesla, like don't go yellow is probably my my my simple.
SPEAKER_01It's all about building that trust and automating when you have that trust versus right.
SPEAKER_02So the tools are emerging, right? So the tools will help you out. We'll have you we'll have you uh kind of covered, uh right, like uh Tesla SNCC and other tools that in in this space. Um but in if very often like the tools are not there yet, right? Like it's not easy, for example, to run as like a fully sandboxed agent today, right? So just uh make sure that you you you don't go fully yellow on everything you do and you don't fully kind of just just just think like don't outsource thinking to your agent, just think for yourself what you're doing, right? That's uh and and use the tools uh as as much as possible.
SPEAKER_01Amazing. Chris, thank you very much. It's been a pleasure to be a pleasure to chat and um and thank you all for listening. Thank you for the questions. Um if you wanted to learn more uh about the integration between Tesla and Sneak, do go to Tesla.io forward slash registry, and from there you'll be able to see a whole ton of skills that automatically have um Sneak running on uh and you'll be able to see the security guidelines and and instructions as to as to uh you know what is it about the e each skill which is potentially causing uh complications.
SPEAKER_01Uh do also go to sneak.io um and check out their scanners there as well. And also um stay tuned in the next few days to see that GA um product release. Thank you. Thank you very much. And uh bye for now.
SPEAKER_00The AI Native Dev is brought to you by TESL, the package manager for skills and context. Your hosts are Guy Pajani and me, Simon Maple. Our producer is Tom Dowler. The AI Native Dev is not just a podcast, it's a community. And we host monthly meetups at the TESL offices in central London. Visit TESL.io forward slash community to learn more, and I hope to see you there.
Developers are about to write skills instead of code, and Microsoft's Daniel Meppiel argues the software factory that runs them needs versioning, a bill of materials and signed receipts before enterprises will trust it....
Paul Stack helped build Terraform. Now he says he'll never write another line of infrastructure as code, because an agent harness runs his team's production infrastructure instead. Watch all the AI DevCon NYC 2026...
Automated software delivery is moving faster than the people who used to approve every release. LaunchDarkly co-founder and CEO Edith Harbaugh tells Guy Podjarny why rollback should be autonomous, why one-shot coding...
Agentic coding has outgrown the text box. Maggie Appleton, staff research engineer at GitHub Next, argues that walls of CLI output and AGENTS.md files serve coding agents far better than the humans steering them. She...
AI attackers now move at machine speed, while most security teams still move at meeting speed. Former Robinhood CSO Caleb Sima has seen an AI attacker up close, and he argues the only way to keep up is to take humans...