Blog
RaggieCode Blog
Technical deep dives into AI coding agents, context-window management, structured session handover, and what it takes to build production-grade code intelligence that actually understands your codebase.
I Stopped Compacting Coding-Agent Conversations
and Started Handing Work to a Fresh Agent
Why context compaction fails for coding tasks. A technical deep dive into structured session handover: what gets lost, how the 12-field schema preserves it, and how a fresh agent resumes exactly where the last one left off. Covers decision rationale, failed approaches, error messages, user constraints, and goal drift across multiple handovers.
Context Compaction vs.
Agent Handover for Long Coding Tasks
A balanced technical comparison of context compaction and structured session handover for AI coding agents. Evaluated across five dimensions: continuity, token cost, correctness, failure modes, and implementation complexity. Includes a decision framework for choosing the right strategy and a look at hybrid approaches.
Quick Links
Direct access to every article and supporting resources.
About This Blog
These articles document the engineering decisions behind RaggieCode .; an open-source AI coding agent that builds a semantic code index with tree-sitter, navigates call graphs with cycle detection, and performs automatic context handover when session limits are reached. Each post covers a specific problem, the approaches considered, the tradeoffs involved, and why the final design took the shape it did.
2
Published articles
12
Handover schema fields
5
Comparison dimensions
Stay Updated
New articles are published as RaggieCode evolves. Follow the project on GitHub for releases, or check back here for deep dives into code intelligence, agent design, and context-window management.