←
Abhinav Asthana3,245 posts
Abhinav Asthana avatar✓

Abhinav Asthana ✓

@a85

CEO and Founder, Postman (Postman.ai)

◷ Joined March 2008▧ 84 Photos and videos
Abhinav Asthana avatar
Abhinav Asthana@a85·Sep 30

.@getpostman is now natively available in your favorite coding agent! The new Postman CLI is designed to bring API engineering for agents: 1️⃣ Works in Git and your code repo - Just drop Postman where your agent lives and works. 2️⃣ No GUI - Work in your coding agent. Bring up the Postman UI for review and approval. 3️⃣ No sign-up - Get started right away. Sign-up when you have to save your work and want more. Install once, work in your agent. Run npm install -g postman-cli, then postman init. This loads Postman's skills into your coding agent so it can create collections, write tests, and much more. Available as a plugin for all coding agents! The skills cover the API lifecycle: find APIs your org already has, generate tests that catch regressions, mock an API before it exists, watch live endpoints for downtime and drift, run checks in CI on every commit, and run performance tests. You can also extend the abilities of your agent with the Context Graph. An agent often sees only the code that is accessible to them. The Context Graph API gives it a map of your team's infrastructure, APIs, and services, and how they depend on each other. As your agent works, you stay the reviewer. Your agent does the work from the command line. Open Postman when you want to review or approve it. Check out more at postman.com!

0:37
Abhinav Asthana avatar
Abhinav Asthana@a85·9h

As we are attributing all productivity growth to AI, also time to add economic loss to the equation with these outages.

Jeff Stein avatar
⟳ Abhinav Asthana retweeted
Jeff Stein@jstein_sun·12h

NEWS: Wall Street execs and NSA have told Congress that outages were caused by patching vulnerabilities identified by AI models, sources say AI did not cause outages but found so many issues that the rush to fix them before an attack overwhelmed systems washingtonsun.com/technology…

Ruben Casas 🦊 avatar
⟳ Abhinav Asthana retweeted
Ruben Casas 🦊@Infoxicador·Nov 16

MCP Events are so cool! inspired by @Poita_ I got ChatGPT on the left playing a chess match against Opus on the right (via Postman Agent Mode) Using MCP Apps and events both agents subscribe to the game and automatically awake when it is their turn... let's see who wins!

1:38
Abhinav Asthana avatar
Abhinav Asthana@a85·Nov 16

Today we’re making Fern the official home for API Developer Portals and SDKs at Postman and announcing a special $3000/year plan available for free for Postman customers to publish their first developer portal on Fern! More than 500,000 organizations, including 98% of the Fortune 500, use Postman to design, build, and test their APIs. Once an API ships, the work moves to the developers who integrate with it. They need accurate documentation and client libraries in the language they already use. Postman’s built-in documentation publishing has helped teams get their docs online for years. Fern builds upon that, purpose-built to serve both human developers and modern AI agents. Rather than split our investment across two docs products, we’re putting our entire engineering team behind Fern, so every improvement lands in one place. Read more here: blog.postman.com/fern-is-pos…

Postman avatar
⟳ Abhinav Asthana retweeted
Postman@getpostman·Nov 16

Claude Code is now an API Engineer. Keep its gate out of reach. Agents will loosen a failing check. With the Postman plugin: - CI runs spec lint + ai-readiness - CLAUDE.md says never edit the gate or lower --min-score Guide: blog.postman.com/postman-plu…

Post attachment
Abhinav Asthana avatar
Abhinav Asthana@a85·Nov 14

Coding agents leaking credentials as a problem is hiding in plain sight. Credentials live in .env files and are copy/pasted with abandon into chats - often at the recommendation of the agent itself.

Abhinav Asthana avatar
Abhinav Asthana@a85·Nov 11

MCP in @getpostman just got a big upgrade. MCP requests in Postman now support scripts and tests. You can set up a connection, check what a server exposes, and test every tool call, prompt and resource read, then run it all in the Collection Runner next to your HTTP requests. Great work by @Infoxicador and team. cc. @dsp_

0:27
OrcaRouter 🐳 avatar
⟳ Abhinav Asthana retweeted
OrcaRouter 🐳@OrcaRouter·Nov 11

Our investigations into the recent wave of AI inference infra attacks point to a concerning pattern. We believe Anthropic’s crackdown on distillation may be accelerating a shift toward stolen credentials and compromised inference infrastructure. Our findings suggest certain Chinese operators are involved in networks reselling frontier model access through heavily discounted AI gateways. The economics are simple: steal API keys, hijack inference capacity, and resell tokens at a fraction of legitimate prices. We suspect these are not isolated incidents, but part of a growing underground economy built around stolen AI compute. Inference infrastructure is becoming a prime target for organized cybercrime.

Abhinav Asthana avatar
Abhinav Asthana@a85·Nov 11

We unfortunately burnt >$1M in 2-3 days on farming for tokens in our trial program and had to shut down model access.

Post attachment
Abhinav Asthana avatar
Abhinav Asthana@a85·Nov 10

Agents are reading docs more than humans! The Latest State of Docs report from @buildwithfern is out. We pulled nine months of traffic across every docs site we host. Between January and September: → Agent requests grew 6x! → 74% of API reference and changelog views came from agents → 93% of agent requests went to Markdown, not HTML We also ranked what to fix first to make your docs better for agents in State of Docs: buildwithfern.com/post/bigge…

bebu avatar
⟳ Abhinav Asthana retweeted
bebu@bebuu_ai·Nov 7

A pretty serious sandbox hole was reported in `anthropics/claude-code` (`#108071`) today on v2.1.319 for Linux. Most developers use the `denyRead` setting (`deny: Read(.env*)`) so Claude Code can never read secret API keys in `.env` files. On Linux, it turns out the sandbox only blocks `.env` files that already existed at the exact millisecond a bash command starts running. If the agent starts a background script, test watcher, or dev server, and a new `.env` file gets created while that command is still running (like from `git pull` or a deploy script), the protection completely misses it and the agent can read the entire secret file in plain text. If you rely on `denyRead` to keep production keys safe on Linux, definitely watch out for long-running background commands until they patch this. Issue: github.com/anthropics/claude…

claire vo 🖤 avatar
⟳ Abhinav Asthana retweeted
claire vo 🖤@clairevo·Nov 7

After working w lots of real eng orgs at @cxodev here’s my current take on software factories - first of all: you do you. If a system works for you, no one is bullying you into adopting something else - model choice & cost optimization is a big driver of “software factory” projects - you should get clear on when you need an agent, when you need a factory, and when you need determinism and validation - good factories eval and measure their way into a structured process that outputs high quality code *for their company/product* - this structured process is usually something like a plan, relevant code discovery, validation, PR/checks/review/scoring, and a release process sometimes even to market facing teams - factories also have non-code objectives, like moving more building into public, making multiplayer more sensical, and enforcing a desired SDLC that would be hard and annoying to enforce organically - ime factories are very good at raising the floor (often necessary in large orgs) but they’re not necessarily going to raise the ceiling for someone like @poteto - that doesn’t make them fake, it just makes them a different tool, and often one needed for whole org transformation - that being said, many fall in love with the idea of “factory” before addressing more fundamental gaps/oppty in their AI Eng stack: everything from repo readiness to fast lints/checks/previews to background agents/automations and basic enablement on AI coding tools/models - also many orgs give someone a factory project and they end up trying to compete with Claude code/codex/whatever instead of designing something unique to their org w specific problems solved really well - which is why I tell people usually to not start w a software factory, but rather w a big factory or a security factory or a UI polish factory. More well scoped and easier to eval - some of this across both factory and non factory setups is pure f*ing theater and token slinging and you should watch out for that (in your company and online) I think @zachlloydtweets writes thoughtfully about factory concepts, and I particularly like how he measures & evals every component, encourages ensuring your factory delivers quant and qual value, and acknowledges the culture/ops benefits of factories like more transparency and XFN learning ymmv but this is what I see! and fwiw I’m trying to build a “factory” for a smaller repo to see if I see the cost/value benefit. So far, no, but those max subs are hard to beat for a small team 😎

Abhinav Asthana avatar
Abhinav Asthana@a85·Nov 6

As capital becomes more expensive and quality of a business becomes more important, AI startups might find out the harsh way that their ARR is neither annual and not in fact revenue.