Bap

Building the AI native Dev community 👋

[Liran Tal @liran_tal]

if you find yourself creating skills a lot, you might want to check out my skills-creator-extra on the @tessl_io registry

[Bap @FernandezBap]

@rseroter @_philschmid great piece - same thread is happening with anthropic with eval support into skill-creator. good signal overall ecosystem is heading the right way. wrote up couple of thoughts if useful for folks https://t.co/il5n05a5w9

[Bap @FernandezBap]

@hwchase17 Fantastic write-up! Great to see LangChain is working on this. Anthropic also shipped eval support into skill-creator, which is a good signal that the overall ecosystem is heading this way! I wrote up my thoughts on this https://t.co/7rZra0FZKa Keen to hear folks' feedback!

[Bap @FernandezBap]

@_philschmid Fantastic writeup! Great to see DeepMind is pausing this. Anthropic also shipped eval support into skill-creator, which is a good signal that the overall ecosystem is heading this way! I wrote up my thoughts on this https://t.co/7rZra0FZKa Keen to hear folks' feedback!

[Guy Podjarny @guypod]

@HashiCorp just released Agent Skills for Terraform and Packer.

What’s interesting here isn’t the idea of Skills itself, but the focus on making agent behaviour repeatable and evaluable. These aren’t prompts, they’re structured, testable bundles of domain knowledge.

We worked with HashiCorp to evaluate the Skills using real task runs and evals in Tessl, looking at where they actually improve outcomes and where they don’t.

Feels like a solid step toward more reliable, AI-native infrastructure workflows. Less guesswork, more verification.

Check it out here: https://t.co/sqBrPeiJ6M

[Bap @FernandezBap]

@ainativedev Great chance to meet the creator of https://t.co/8f0tFLsL36 🙏

[Bap @FernandezBap]

Thought Codex got better.. 🫠

[Lucas Atkins @latkins]

I be using Codex, as you can tell.

[Bap @FernandezBap]

Traceability is becoming more important as AI systems scale, and graphs definitely make data flows clearer.

[AI Native Dev @ainativedev]

Most devs discard structure just to fit into vector search.

Michael Hunger (@mesirii) and his team at @neo4j tackle this for 80+ Fortune 100 companies by storing data as connected nodes, preserving relationships that help you make smarter decisions.

On this episode of AI Native Dev, Michael Hunger joins @sjmaple to explore how graph databases reshape AI workflows and how MCP servers wrap APIs to enable context aware developer experiences.

Catch their full conversation now.

[AI Native Dev @ainativedev]

🚨 New best open-source model for agentic coding?

Devstral (by @MistralAI & @allhands_ai) just hit 46.8% on SWE-Bench Verified, topping all other open weights models.
- Low hardware requirements
- Open weights
- Test-loop agent focus

@FernandezBap breaks down the news for us ➡️ https://t.co/fig84RfEVY

#Mistral #AINativeDev #AgenticCoding

[Simon Maple @sjmaple]

I recently sat down with Anton Arhipov from JetBrains to talk about IntelliJ IDEA’s shift toward becoming an AI-powered IDE.

He walked me through some of the intelligent plugins, like context-aware code suggestions and machine learning–driven refactoring, that are fundamentally improving code accuracy and accelerating development cycles.

If you're curious where IDEs and developer tools are headed, links to the full episode are in thread.

[Bap @FernandezBap]

Running on consumer hardware is especially exciting because it pushes forward the accessibility and democratization of powerful models.

[AI Native Dev @ainativedev]

🚨 New best open-source model for agentic coding?

Devstral (by @MistralAI & @allhands_ai) just hit 46.8% on SWE-Bench Verified, topping all other open weights models.
- Low hardware requirements
- Open weights
- Test-loop agent focus

@FernandezBap breaks down the news for us ➡️ https://t.co/fig84RfEVY

#Mistral #AINativeDev #AgenticCoding

[Dion Almaer @dalmaer]

I use playwright-mcp to drive the browser, but why not go directly to CDP to get full access to networking and performance as well as DOM etc?

@autoconfig wired this up with devtools/mcp! Love it!

[Bap @FernandezBap]

This is super exciting stuff.

[AI Native Dev @ainativedev]

@github Copilot just leveled up! A cloud-based agent that opens PRs, writes code, and responds to reviews autonomously.

Not autocomplete. Task-level execution in the cloud.
CI-safe, branch-safe and thoughtfully designed.

Read our breakdown:
🔗 https://t.co/XwMWCT63AG

#Build2025 #AIdev #GitHubCopilot

[AI Native Dev @ainativedev]

JVM devs, this one’s for you: @springrod just dropped an OSS agent framework.

JVM-native, Kotlin-based, and focused on agentic flows mixing LLM-prompted interactions with code and domain models.

@dalmaer got hands-on and shared his thoughts here → https://t.co/nJYUdVTI8b

Curious what the JVM crowd makes of this?

#Java #Kotlin #AINativeDev #News #OSSXL

[Bap @FernandezBap]

@ainativedev @springrod @dalmaer Refreshing to see agent frameworks go beyond Python/JS/TS. JVM deserves solid tooling.

[Bap @FernandezBap]

@ainativedev @sjmaple On that same thread, I'd also recommend the Firecrawl and Obsidian MCP.

[Bap @FernandezBap]

On that same thread, I'd also recommend the Firecrawl and Obsidian MCP.

[AI Native Dev @ainativedev]

12 MCP Servers That Will Make You A More Productive Developer ⬇️

MCP servers are like APIs for your agent’s brain. You define the tools. The agent learns to use them. And suddenly, your LLM can run a Docker container or file a GitHub issue.

@sjmaple breaks down 12 MCP servers you should start tinkering with. Links in the comments.

#MCP #GitHub #Docker #AINativeDev

[AI Native Dev @ainativedev]

“2025 is the year of agents,” says @gdb. In line with that, @OpenAI has reintroduced a revamped version of Codex with Codex Agent. @dalmaer dives into the details and shares his thoughts: https://t.co/Imb57LZz8V

#AINativeDevelopment #AIForDevelopers #CLI #Codex #ClaudeCode

[AI Native Dev @ainativedev]

@Sourcegraph's Amp is now available to all. Built for agentic coding, Amp is designed to handle complex, multi-step development workflows.

@thorstenball notes: “Amp now writes 70–80% of the code I commit.” @dalmaer comments on this release: https://t.co/RXXI74QdsF

#AINativeDev #DevTools #AIProgramming #VSCode #AgenticCoding

[Bap @FernandezBap]

@ainativedev @Sourcegraph @thorstenball @dalmaer Exciting. Adding it to the weekend list of tools to tinker with. :)