What Codag does
Codag is a developer tool that compresses large volumes of infrastructure logs into compact, structured JSON summaries designed for AI agents to process within token budgets. It transforms log files spanning hundreds of megabytes and over a million lines into small, schema-valid outputs that retain only diagnostically relevant entries.
Key capabilities
Codag preserves factual accuracy by citing the real line number of every kept log entry and tagging it by role such as root cause, trigger, or consequence, rather than generating lossy summaries. It supports JSON logs, syslog, Kubernetes, Docker, AWS CloudWatch, and unstructured application logs, and includes preprocessing for redacting sensitive data. It integrates with AI coding agents such as Claude Code and Codex through the Model Context Protocol or HTTPS wrappers.
Who it's for
Codag is aimed at AI coding agents, platform and infrastructure teams, and individual developers who need to debug issues across large logs while reducing token consumption.