Blog
Notes from the workbench. What we're building, how we built it, what we learned.
Orchestack: Multi-tenant Orchestration for Agent Systems
What Orchestack is, how it's built, and what it integrates with. The orchestration layer for everything else CDR ships.
cdr-home: A Website for Humans and Agents at the Same Time
How this site serves the same content as HTML for people and as JSON or Markdown for agents, without compromising either side.
CDRcache: Hash-Addressed Caching for Agent Pipelines
If an agent produces the same output for the same input, you should only have to run it once. CDRcache is the layer that makes that true.
mae: A Minimal Agentic Environment for Linux Maintenance
Most Linux maintenance is repetitive and error-prone. mae is the persistent agent we built to take over the toil.
rlm-linux: Recursive Orchestration for Fedora-Based Systems
Build and maintain your distro from first principles with a Sakana-style RL conductor over a tiered pool of local and remote LLM workers.
Pachyterm: A GPU-Accelerated Terminal with Native AI Integration
Sub-50ms time-to-first-frame, POSIX-compliant, and a `p` prefix that turns the rest of any line into an LLM prompt.
CDRbrowser: A Clean-Slate Browser Built and Maintained by AI
Not a Chromium fork. Not a Firefox fork. A Rust web engine paired with a native AI synthesis plane, with the codebase itself maintained by AI.
TopoLI: Topological Pruning for ColBERT-Style Retrieval
ColBERT stores documents as bags of token embeddings. Most of those tokens don't matter for retrieval. TopoLI uses persistent homology to find the ones that do.
fpre: A First Principles Reasoning Engine
Pure LLM reasoning hits a ceiling on problems that need real symbolic structure. fpre is a tight loop between a model that proposes and a typed engine that verifies.
CDRmem: Memory Access Gates and Fast-Weight Memory for Language Models
Less about storing vectors and more about how a model accesses what it remembers. Memory gates, fast-weight modules, and deterministic gating.
CDRdistill: H-Neuron Detection for Hallucination-Aware Distillation
Treating hallucination as something locatable in the network, not a black-box behavior. The result: smaller models that are measurably more grounded.