AI AGENT / RESEARCH AUTOMATION / TOOL-USING SYSTEMS

Building agents that think with tools.

I explore agentic workflows, context engineering, evaluation loops, and the interfaces that make human-AI collaboration feel precise instead of mysterious.

$ boot agent --profile=research

View work GitHub

36 agent runs

128 notes indexed

07 experiments active

1.0 public interface

Agent Stack

Systems, not just prompts.

planner / executor / reviewer

Agent Systems

Designing workflows where language models plan, call tools, inspect results, and recover from messy states.

search / ingest / synthesize

Research Automation

Turning papers, docs, and notes into structured context that can be searched, tested, and reused.

trace / grade / iterate

Evaluation Loops

Building feedback paths for prompts, agents, and generated artifacts so improvements are visible.

Run Logs

A visible trail of experiments.

09:42 indexed agent-runtime notes ok

10:18 validated tool-call contract ok

11:03 queued evaluation harness wait

13:27 published research digest ship

Work

Where the experiments collect.

Contact

Open a channel.

GitHub Email