Diego Comas retweeted
Agentic Batch Changes is available today.
Describe a change once. It runs across 10 or 10,000 repositories, adapts where they differ, and reacts when CI fails.
Agentic Batch Changes uses outcome-based pricing. You pay per changeset merged into your codebase.
If a pull request doesn't merge, you don't pay for it.
Read more here: sourcegraph.com/blog/introdu…
Diego Comas retweeted
Today we're launching Agentic Batch Changes in public beta, and we're proud of how it's turning out.
Mercari engineer Patrick Klitzke used Agentic Batch Changes during our experimental preview to scope and begin patching a GitHub Actions security vulnerability across their codebase.
"I was able to fix it with one prompt on both the Help Center frontend and backend, then extended this to all repos in Mercari. I found around 80 potential repos affected."
The thing that made this work wasn't speed. It was the agent's ability to reason about each repository individually. Mercari's repos have similar setups, but not the same setup. A scripted find-and-replace would have collapsed. The agent handled the variation per repo, reacted to CI, and pushed follow-up commits where needed.
Canva has been in the preview too, using it to split Batch Changes by code ownership across a Bazel monorepo.
This is a public beta. There are rough edges and we're honest about that. But the core capability is working in real customer environments and we're ready to put it in more hands.
Available today for Sourcegraph Cloud customers. Free during the beta period. Self-hosted arrives July 8 with Sourcegraph 7.5.
Full announcement: sourcegraph.com/agentic-batc…
Diego Comas retweeted
Today, we're announcing Sourcegraph 7.0, a release that marks the beginning of a new chapter for our company and product.
Over the past several releases, we've shifted our focus. We're doubling down on being the intelligence layer that developers and AI agents rely on to navigate, understand, and operate on large codebases.
sourcegraph.com/blog/a-new-e…
Diego Comas retweeted
Big news! Sourcegraph & @AmpCode are becoming two separate companies.
Read why:
sourcegraph.com/blog/why-sou…
I use Deep Search all the time, and it helps my teams onboard faster, investigate issues, and find what we need across the codebase in seconds. Huge productivity boost.
Diego Comas retweeted
We made Amp Free.
It's powered by great tokens and tasteful ads.
Agentic coding is now free for everyone.
Add this to your workflow! Early detection of dependency issues will definitely reduce headaches.
Diego Comas retweeted
We flipped the switch: Amp Tab is now on by default for new installs.
It’s our free completion engine for manual edits in VS Code. It’s fast, knows your recent changes and compiler errors, and suggests cross-file updates when needed.
Already on Amp? Run `Enable Amp Tab` from cmd/ctrl+shift+p.
VS Code blocked this malicious extension quickly, but in an alternative marketplace (Open VSX, used by forks like Cursor/Windsurf) it still hit 54K+ installs.
A fake “Solidity Language” highlighter, it was malware stealing $500K in crypto.
Vetting extensions and their sources is just as critical as securing the code we write.
This extension never made it into @code because our marketplace team banned it about 2 seconds after it was published.
Not in our house.
securelist.com/open-source-p…
Diego Comas retweeted
We’ve been working with OpenAI over the last few weeks on GPT-5 in Amp.
We’re still evaluating where and how to use GPT-5 (after all, it took 6 months to figure out 3.5 Sonnet’s true capabilities).
Join us and try GPT-5 in Amp now:
Diego Comas retweeted
In this episode, Beyang and Thorsten discuss strategies for effective agentic coding, including the 101 of how it's different from coding with chat LLMs, the key constraint of the context window, how and where subagents can help, and the new oracle subagent which combines multiple LLMs.
00:38 Intros
03:20 How coding with agents is very different from coding with prior AI tools
10:31 Example: fix a simple issue
14:13 Example: debugging an issue with an MCP server
21:50 Example: unifying two build scripts
25:09 How the context window is a key constraint
31:01 Why it's best to focus on one thing at a time
33:09 Subagents and context windows
33:49 The codebase search subagent
38:33 General-purpose subagents
44:05 When to use subagents
46:44 The oracle subagent and o3
51:32 Multi-model agents
Diego Comas retweeted
Heard from someone who found a critical vuln in a popular site using Amp, and another who fought off a DDoS attack with Amp’s help.
We want to make Amp great for security and incident response, not just coding: Amp optimizes for quality (not cost), never locks you out at critical times, and has robust (automatic) fallback to Vertex for Claude inference in case of Anthropic downtime, when you might need it most.
What can we do to make Amp better for security and ops/SRE stuff?
Excited to join the upcoming State of DevSecOps 2025 livestream hosted by @christophetd and organized by @datadoghq, with Clea Ostendorf and Fred Blaise.
Learn the latest insights on modern DevSecOps practices pulled from real-world organizations.
At @Sourcegraph , we have seen how these challenges evolve when you are building AI-powered tools like @AmpCode and supporting a fast-moving global engineering team.
I am looking forward to sharing what has worked for us and hearing from others in the thick of it too.
📅 Live on July 15
🔗 Register here : datadoghq.com/event/devsecop…
. @AmpCode is now SOC 2 Type II certified! ✅
Enterprise-grade security from the start — zero exceptions, strong controls, fully audit-ready. 🛡️
This adds to @Sourcegraph's ISO 27001, GDPR, and CCPA compliance.
We’re building secure, scalable AI from day one. 🔐
Learn more → ampcode.com/security
Diego Comas retweeted
We feel this is the best way to take advantage of o3's powerful reasoning capabilities within your agentic coding workflow. It's not an agnostic model selector, but an intentional application of each model to what it does best.
