01 / Featured tools

These projects support technical reasoning, repeatable execution, evidence capture, and honest failure states. Maturity and limitations are shown alongside capability.

01
Engineering orchestration / CFD

Aero

An experimental engineering agent for planning, teaching, executing, and reviewing CFD workflows across first-order methods, OpenFOAM, and supervised Fluent workflows.

  • Engineering problem framing and fidelity selection
  • Geometry, mesh, boundary-condition, and preflight checks
  • Convergence, conservation, and validation gates
Active development / experimental2026View technical review
02
Bounded agent execution

ForgeClaw

A framework for supervised agent execution with explicit authority boundaries, observable tool handoffs, evidence capture, verification, and controlled escalation.

  • Declared execution surfaces and explicit tool authority
  • Bounded action loops with policy checks
  • Evidence capture and concrete failure classification
Alpha / experimentalv0.1.0-alpha.6View technical review
03
Evidence-aware document processing

OCR97

A local document and image-text processing tool for screenshots, scans, forms, and image-based PDFs with adaptive preprocessing, lane selection, uncertainty visibility, and inspectable output.

  • Native PDF text and local OCR routing
  • Adaptive image preprocessing
  • Screenshots, scans, forms, and image-based PDFs
Alpha / experimentalv0.1.0-alphaView technical review
02 / Engineering software

Aero connects software work directly to engineering practice.

Aero connects engineering problem framing, fidelity selection, solver setup, execution evidence, and review without treating an AI-generated answer as computational authority.

03 / Local AI & automation

Bounded execution and evidence-aware extraction.

ForgeClaw and OCR97 focus on the parts of automation that are easy to hide: authority, uncertainty, evidence quality, verification, and failure.

01
Bounded agent execution

ForgeClaw

ForgeClaw explores how an agent can act on a declared surface, preserve what happened, and require separate evidence before a result is accepted.

Alpha / experimentalBoundaries & evidence
02
Evidence-aware document processing

OCR97

OCR97 treats extracted text as evidence with provenance and uncertainty rather than as a guaranteed transcription ready for silent downstream use.

Alpha / experimentalBoundaries & evidence
04 / Research notes & architecture

The reasoning behind the systems.

Selected notes connect implementation decisions to engineering concerns: defined authority, observable execution, independent acceptance, and evidence that survives failure.

Why OCR Quality Matters to Agentic Workflows

A draft technical note on how missing rows, damaged labels, and uncertain text propagate into downstream tool decisions.

Draft
Measuring Local AI Savings Without Lying to Yourself

A draft measurement template for comparing local execution, cloud planning, latency, energy, and failure costs without cherry-picking.

Draft
05 / GitHub

Public source and project records.

ForgeClaw and OCR97 were originally published through Precision Arts Lab. That attribution is secondary here to their role in Alex's independent technical work.