MCP went the HTTP way: what changed in the 2026 protocol
MCP's 2026-07-28 revision removes sessions and makes each request self-contained. Here is what changed and how to use it.
K
Kirti Rathore
8 min read
Latest Articles
Insights on AI-powered bug fixing, automated debugging, and modern software engineering.
Building For Agents
A practical guide for engineers designing APIs, MCP servers, observability systems, or delegated services for AI agents.
K
Kirti Rathore
13 min read
Building AI SRE In-House: Costs and Tradeoffs to Consider
What engineering leaders should model before committing to build, operate, and continuously improve AI SRE in-house.
R
Robin Singh
5 min read
The $180K Debugging Tax: What Production Incidents Really Cost Your Engineering Team
A transparent calculator for the hidden cost of production incident investigation - and the engineering capacity AI debugging agents could return.
R
Robin Singh
5 min read
How to build your context API
A practical guide to building an LLM-facing context API.
K
Kirti Rathore
10 min read
Runtime bugs: when the first 30 minutes are critical
A stuck I/O incident looked invisible in logs, metrics, and traces. The useful evidence came from reproducing the hang, forcing a core dump, and looking at it in GDB.
K
Kirti Rathore
5 min read
Fixing bugs reduces to having their reproduction test case.
A root cause is only useful once it can be reproduced. The practical debugging loop is to localize the fault, turn the alert into a reproduction, and test hypotheses until the fix is validated.
K
Kirti Rathore
4 min read
What if every bug report came with root cause attached?
Most alert queues are full of duplicates, expired assumptions, and evidence-free mysteries. The useful AI workflow is not just RCA. It is hypothesis validation all the way to a proven fix.
K
Kirti Rathore
5 min read
FixBugs is now live on the GitHub Marketplace!
The FixBugs GitHub App is now available on GitHub Marketplace. Install fixbugs-app on a repo, start from a real issue, and get hypotheses, reproductions, fixes, PRs, and validation evidence inside GitHub.
K
Kirti Rathore
7 min read
High-performance AI agents are distributed systems
A technical case study on making FixBugs faster by treating LLM agents like distributed systems: chunking, streaming, queues, token math, and concurrency limits.
K
Kirti Rathore
11 min read
Oncall isn't supposed to be this hard
Well-configured alerts connect the symptom to traces, logs, deploys, and the suspect commit.
K
Kirti Rathore
6 min read
FixBugs is not a GitHub bot
Most people who land on fixbugs.ai still think we fix GitHub issues. That framing is wrong, and it's our fault. This is where debugging actually happens — the IDE, the issue tracker, the observability pipe — and what a debugging agent has to see to be useful.
K
Kirti Rathore
7 min read
Engineering approach: Startup Mode v/s Big Tech Mode
Discover why Big Tech performance engineering practices can destroy startup velocity. Learn when to optimize and when to move fast, with real examples from Figma and Stripe.
K
Kirti Rathore
7 min read
How To Build High Performance AI Agents - Insights from PyDelhi Talk
Key insights and strategies from my PyDelhi talk on building high-performance LLM agents that can process gigabytes of data and deliver results in under a minute, with lessons learned from optimizing FixBugs.
K
Kirti Rathore
2 min read
FixBugs Web
Try the FixBugs auto-triage and fix tool right from your browser!
K
Kirti Rathore
3 min read