indexing the chaos
built graph-based workflow systems for a product saas platform: architected nested node graphs with cycle prevention and consistent layout, implemented singleton memory model and partial updates for performance, built ai-assisted editing with vercel ai sdk for chatbot-driven workflow construction, designed extensible plugin architecture, led webpack to rspack migration (~25% faster builds).
It would be a minor statement to say that it fucked up. It would straight on jump to coding without even looking or exploring the codebase. Okay that was my fault because I did not give it proper system prompts. All right let's give it a system prompt.
Can a tiny model like solve a simple embodied text game if given access to a simulated environment? > Note: This is not an official ALFWorld task but a simplified version for testing purposes. put the apple in the…
Shiryouku is a research workspace for turning messy source material into searchable, citeable context. It ingests uploaded documents and live website pages, splits them into linked chunks, indexes them with both keyword…
An AI brain that combines zenrag and zenpreference for knowledge retrieval and preference recall. Includes an MCP server and CLI. - Unified context — Combines knowledge (RAG) and preferences in one call - MCP server —…
A semantic preference store for AI applications. Store key-value preferences with natural language content and retrieve them by semantic similarity. - Semantic retrieval — Find preferences by meaning, not just exact…
A minimal RAG (Retrieval Augmented Generation) library for the Vercel AI SDK. Semantic search over documents with SQLite or in-memory storage. - Chunking — Configurable text chunking with overlap - Vector stores —…
Select context for coding agents directly from OpenTUI apps. Click any component in your terminal UI to copy its component name, type, and hierarchy to your clipboard—ready to paste into Cursor, Claude Code, Copilot, or…