Public substrate · active harness

Public substrate & the reusable pieces pulled out of a working agent harness.

The private harness I use every day is a local operating layer for agentic engineering work. The public part is smaller and more useful: reusable references for safety boundaries, repeatable lab workflows, provider routing, and durable memory. The goal is not to publish every local wire. It is to expose the parts that another engineer can inspect, run, and adapt without inheriting my machine.

The security posture is deliberate: secret-bearing local glue stays private; public repos are minimal, MIT-licensed references with narrow responsibilities. A visitor should be able to understand the pattern without receiving operational details that belong on one workstation.

The public extracts are the *-mini repos: safe-mini for safety boundaries, lab-mini for data-science labbing, route-mini for multi-provider LLM routing with fallback, and memory-mini for durable agent memory. They are not the whole harness; they are inspectable references for the pieces that should be reusable outside my machine.

Status · active use · public substrate extracted

Public substrate

Public substrate

The *-mini family — minimal, MIT, well-tested public references for the moving parts of an agent harness. safe-mini (safety reference), lab-mini (data-science labbing), route-mini (multi-provider LLM routing with fallback), memory-mini (durable agent memory).

← All work