DroidCTX is an Open Source CLI Tool that generates documents about production infrastructure that can be easily plugged into coding agents like Claude Code and Cursor. This tool makes the agent respond faster, more accurately, and with fewer token consumption because instead of needing to explore all the different MCP servers, it instantly knows where to begin investigations.
Hello everyone!
This is Sid and Dipesh from DrDroid. We started building DroidCTX because we saw one issue happen frequently while using coding agents to debug production issues -- they did a lot many tool calls and queried data across multiple tools trying to explore the production infrastructure.
While working on it, I realised that if I can make agent have access to explore the entire production infrastructure across all tools in one shot, including cross-references, it could significantly accelerate agentic debuggging.
We did that and overnight we saw that more than 30% token reduction and debugging speed caught up when we let agent do initial exploration in that setup.
This motivated us to create DroidCTX -- an easy way in which you can compile your production infrastructure into a set of .MD files that could be relevant to the agent.
We've been able to generate infrastructure docs for our entire company in less than 60s and were able to generate the same for < $5B+ enterprises in less than 2 hours.
Report
Congrats on the launch! 🚀 This solves a problem I've felt firsthand — AI coding agents like Claude Code are incredibly powerful but they're essentially "flying blind" when it comes to your actual production infrastructure. They know your code but they don't know that your Kubernetes pod is OOMKilling or that a specific Grafana dashboard is showing elevated error rates.
The idea of giving the agent a structured Markdown snapshot of your real infrastructure state before debugging starts is genuinely clever. It's the difference between asking a consultant who's read your docs vs. one who's actually looked at your dashboards.
Quick question for the team: How do you handle sensitive credentials or PII that might appear in infrastructure metadata when syncing from tools like Datadog or AWS? Is there a built-in scrubbing/masking layer before the context file is written? That seems critical for enterprise adoption. 🛡️
Report
No reviews yetBe the first to leave a review for DroidCTX
DrDroid
Congrats on the launch! 🚀 This solves a problem I've felt firsthand — AI coding agents like Claude Code are incredibly powerful but they're essentially "flying blind" when it comes to your actual production infrastructure. They know your code but they don't know that your Kubernetes pod is OOMKilling or that a specific Grafana dashboard is showing elevated error rates.
The idea of giving the agent a structured Markdown snapshot of your real infrastructure state before debugging starts is genuinely clever. It's the difference between asking a consultant who's read your docs vs. one who's actually looked at your dashboards.
Quick question for the team: How do you handle sensitive credentials or PII that might appear in infrastructure metadata when syncing from tools like Datadog or AWS? Is there a built-in scrubbing/masking layer before the context file is written? That seems critical for enterprise adoption. 🛡️