**`MAINTAINER_NOTES.md`**
```markdown
# Maintainer Notes & Technical Obstacles

## Technical & System Constraints Identified
1. **Sandbox Environment Isolation:** AI execution environments operate in air-gapped Linux containers without external network access or direct cloud storage mounting. File scanning scripts requiring external asset access (such as OpenCV signature matching against remote drives) must run locally on user hardware.
2. **Template Matching Scale Sensitivity:** OpenCV `matchTemplate` functions are not scale-invariant. If signature or phase-space patterns differ in image dimensions, standard matching fails unless executed through a multi-scale image pyramid loop.
3. **Session Context Truncation:** Token-heavy conversational sessions containing image attachments trigger automated context dropping. Core mathematical models, proofs, and source code must be persisted immediately to Git version control.
4. **Symbolic Geometric Cancellation:** Standard quantum action uses $\hbar = \frac{h}{2\pi}$ as a cyclic-to-angular conversion factor. Replacing scalar $h$ with a scale-dependent fractal density $H(\delta) = 2\pi \cdot f(\delta)$ geometrically absorbs $2\pi$ at non-differentiable phase-space boundaries. This yields scale-invariance at $D_H = 2.0$ ($\hbar = \lambda$) and an inverse action density divergence $\chi(\delta) = \frac{1}{\lambda}\delta^{D_H-2}$ when $D_H > 2.0$.

## Attribution Summary
* **Theoretical Origin & Hypothesis:** Albert Lane
* **Julia Package & Symbolic Verification:** Gemini (Google AI)
* **License:** MIT License