I build tools for AI operator token measurement and governance. Start with the live product, run the local tool, or explore the implementation:
- Use SignalAF / SigRank — leaderboard, personal profiles, Compare, and Hall of Signal.
- Run the CLI or connect an AI client — local token telemetry, terminal dashboard, and MCP tools.
- Explore the web app source — application, public API, and signed submission flow.
npx sigrank
MO§ES™ governs. Upsilon measures. SigRank publishes the public leaderboard and proof surface on SignalAF. TTEOP defines the protocol semantics.
Measurement uses four token pillars: input, output, cache-write, and cache-read. Prompt text, response text, and source code are not required for ranked telemetry. Signed submissions also carry identity, device/signature, platform, and report metadata; see the submission privacy details.
The current package version is shown in the npm badge above. See the live board for scored operators; directory profiles and registered accounts are separate counts.
Explore AI operator token-processing patterns, calculated from four token counts. The metrics do not establish model quality, work quality, productivity, or business outcomes.
Yield (Υ) = (cache_read × output) / input². The server verifies and re-scores submitted telemetry.
| Surface | URL | What |
|---|---|---|
| Leaderboard | signalaf.com | Public rankings, score calculator, methodology |
| SigEconomy | sigeconomy.com | Competitive operator-ranking arena (Cloudflare Workers) |
| CLI + MCP | npx sigrank |
On-device token scanner, TUI dashboard, 20+ MCP tools |
| VS Code | extension | Inline Yield measurement in your editor |
Interactive coaching agent that reads your token logs, builds a taste profile, and coaches your cascade.
The metabolic panel to SigRank's calorie counter.
Repository · npx signa
Meaning is measurable. A Conservation Law for Commitment in Language — falsifiable, empirically tested, patent pending.
34-paper research program. If a "may" becomes a "shall" during compression, we catch it.
Foundational law · Full program CC-BY-4.0 · Zenodo DOI
Governance framework for AI operator evaluation. Content-free telemetry, provenance labels, developmental measurement.
The architecture is sovereign. The artifacts are licensable. The core that produces them is never for sale.
| Surface | URL | What |
|---|---|---|
| Governance | mos2es.org | Methodology, concepts, AI evaluation pages |
| Enterprise | enterprise.mos2es.org | B2B pilot platform |
| MCP | mcp.mos2es.org | MO§ES™ MCP server |
Governed AI agent marketplace with constitutional governance baked into the HTTP layer.
Multi-homing breaks network effects. The compounding advantage is the non-portable provenance chain.
- fundscore — investor-readiness repo scorer.
- Application Hub — founder-first application infrastructure.
Research, provenance, and public records
| 📄 4 patent filings | 63/877,177 · 63/883,018 · 19/426,028 · 63/991,282 |
| 🏷️ MO§ES™ trademark | TM 99408355, IC 042 |
| 🔬 Zenodo DOIs | 10.5281/zenodo.20029607 · 10.5281/zenodo.19105225 · 10.5281/zenodo.20031715 |
| 🎓 ORCID | 0009-0002-9904-5390 |
| 🏢 Entity | Ello Cello LLC |
| 🌐 Live domains | signalaf.com · sigeconomy.com · mos2es.org · signomy.xyz · mos2es.xyz |
| 🧪 Coherence stress test | 41% baseline → 80–85% measured (5-phase, rival AI architectures) |
| 🎙️ Live demos | KASSA voice AI · Signomy |
| 📰 Public record | Grok stress thread (339 exchanges, 13 days) · Stanford submission |
The record is the moat.
C(T(S)) ≈ C(S) · Υ = (cache_read × output) / input²
GitHub activity
