A specialized browser-based diagnostic tool designed to detect signal corruption in audio waveforms and logical risks in text specifications. It functions as an objective gatekeeper using deterministic consistency checks (as logged).
Core scoring is deterministic and local: the same input yields identical scores. Optional services (e.g., transcription) never affect scoring. Every decision is traceable to versioned rules and measured features recorded in the audit log.
- Audio Scan: Detects clipping, peak overflows, and data dropouts
- Text Scan: Analyzes ambiguity, numerical density, and responsibility gaps
- Privacy-First: Core scoring is local and deterministic; your data remains within the local boundary
- Technical Framework: Implementation of Rule-based consistency gating under fixed constraints
No installation required — a single-file solution for professional risk management.
- Audio Integrity Scanner: Real-time analysis of
clip_rate,true_peak,crest_factor, anddropout_count - Spec/Contract Analyzer: Scans for "Poem Specs" and "Undefined Responsibilities" using weighted keyword density
- Dynamic Calibration: Switch between STRICT, DEFAULT, and LENIENT presets to match project rigidity
- HUD Interface: Futuristic, high-visibility dashboard for immediate decision making
- Audit Logging: Traceable to versioned rules and measured features recorded in the audit log for later export
The Text Analyzer operates by calculating the "Constraint Density" of the input document using rule-based thresholds (versioned) with no ML in scoring. It categorizes keywords into three distinct layers to determine if a specification has a valid "Engineering Constraints."
Targets "weasel words" and non-committal language that obscure technical reality. High density in this category triggers a POEM_SPEC_RISK.
- Detected Patterns:
case-by-case,flexible,optimize,AI-driven,best-effort,at discretion. - Logic: These words often indicate a lack of concrete planning or an attempt to shift responsibility to "as-needed" scenarios, making the path to success defined as a boundary incompleteness (fails consistency gates).
Scans for SI units and technical abbreviations that provide objective measurement.
- Detected Patterns:
ms,dB,%,Hz,hours,SLA,MTBF,throughput,latency. - Logic: A specification without numbers is not engineering; it is a wish. The sensor requires a minimum density of these units to allow a PASS status.
Identifies legal and operational boundaries that define the "In-Scope" and "Out-of-Scope" regions.
- Detected Patterns:
guarantee,scope,exclusion,liability,agreement,specification. - Logic: This layer ensures that numerical targets are backed by clear accountability. It detects the Forward/Reverse predictability gap where one party may be assuming infinite risk without defined exclusions.
- RED (CUT): High Ambiguity + Zero Numerics. The consistency checks fail to converge under current constraints.
- ORANGE (HOLD): Numerics present but no Responsibility defined (or vice-versa). The boundary is incomplete.
- GREEN (PASS): High Numerical density balanced with Consistency / residual gate validation.
👉 https://danger-sensor.dim.productions (Still runs locally in your browser)
Hosting only serves the static file. Analysis runs fully in-browser; no data upload is required.
Drag and drop your files or paste text to initialize the boundary scan.
danger-sensor/
├── index.html # Main HUD application (Single-file)
└── README.md
- Chrome, Safari, Firefox, Edge (Latest versions recommended)
- Audio Support: Standard Web Audio API compatibility
- No external libraries / no scoring API: Pure HTML/JS/CSS implementation; no external libraries or APIs required
For collaboration, technical inquiries, or licensing: info@dim.productions
© 2026 DIMProductions. This tool supports secure data and contract validation workflows.