Spaces:
Running
Running
| a11oy — Warhacker Exhaustive Demos | |
| NOTICE / Attribution | |
| SZL Holdings · 2026-06-06 | |
| ==================================================================== | |
| SOVEREIGNTY STATEMENT | |
| -------------------------------------------------------------------- | |
| The Warhacker demo backend (szl_warhacker_demos.py) and the demo page | |
| (pages/warhacker.html) are SOVEREIGN: | |
| * ZERO runtime third-party dependencies. The backend is pure-Python | |
| using only the standard library (hashlib, json, time, math, base64). | |
| * ZERO runtime CDN. The page ships no external scripts, no external | |
| stylesheets, and no network fonts. It uses a system font stack and | |
| same-origin fetches to /api/a11oy/v1/wh-demo/* only. | |
| * No proprietary code is bundled. Where an established open-source | |
| library implements an algorithm we needed, we REIMPLEMENTED the | |
| well-known algorithm/pattern as our own clean-room code rather than | |
| importing the library, so nothing needs to be vendored at runtime. | |
| Because no third-party source is copied or linked at runtime, this NOTICE | |
| is an ATTRIBUTION OF PATTERNS we learned from — an act of good citizenship, | |
| not a license obligation. All listed projects use permissive (MIT / Apache-2.0 | |
| / BSD / ISC) or open (NASA NOSA) licenses that allow reimplementation. | |
| ALGORITHMS / PATTERNS REIMPLEMENTED (clean-room), BY DEMO | |
| -------------------------------------------------------------------- | |
| 1. CANNONICO — AI oversight for autonomous drones (REAL TODAY) | |
| * STL (Signal Temporal Logic) robustness rho — online-monitor | |
| semantics. Pattern reference: RTAMT (MIT). We compute rho as the | |
| min-over-time, min-over-conjuncts margin to the authorized boundary. | |
| * Geofence containment (ray-cast point-in-polygon, signed distance to | |
| polygon edges). Pattern reference: NASA PolyCARP / DAIDALUS (NASA | |
| Open Source Agreement, NOSA). Reimplemented as our own geometry. | |
| * 13-axis conjunctive ROE gate — policy-as-conjunction. Pattern | |
| reference: Open Policy Agent / Rego (Apache-2.0); our own Lambda | |
| conjunctive-gate doctrine. authorized = AND of all 13 axis predicates. | |
| * Conformal prediction interval (split-conformal). Our own W5-3 / W7-4 | |
| proven construction (coverage + rank-count p-value). Never claims 100%. | |
| 2. TYCHEE — air-gapped ground-software deploy (ROADMAP; substrate REAL) | |
| * OCI image-layer SHA-256 digests + Merkle bundle root (RFC-6962 style). | |
| Pattern reference: sigstore/rekor, in-toto, cosign (all Apache-2.0). | |
| * DSSE envelope (Dead Simple Signing Envelope) PAE. Pattern reference: | |
| secure-systems-lab DSSE spec / in-toto (Apache-2.0). | |
| * Admission-controller "block on digest mismatch". Pattern reference: | |
| Pepr / UDS Core (Apache-2.0); Zarf air-gap packaging (Apache-2.0). | |
| 3. HANGAR2APPS — readiness dashboard + audit (ROADMAP; substrate REAL) | |
| * HL7 FHIR R4 Bundle parse (Immunization / Observation / Flag / Task). | |
| Open standard: HL7 FHIR. Pattern reference: HAPI-FHIR (Apache-2.0). | |
| Sample data only — NO PHI. | |
| * 5-axis conjunctive readiness gate (our Lambda doctrine). | |
| * Append-only signed Task attestation chain (rekor RFC-6962 pattern). | |
| 4. CYBER RTS — trajectory/orbit operational context (ROADMAP; substrate REAL) | |
| * SGP4 mean-element propagation (Kepler 3rd law for semi-major axis; | |
| Kepler-equation solve E - e*sinE = M; perifocal->ECI rotation by | |
| RAAN/inc/argp). Pattern reference: python-sgp4 (Brandon Rhodes, MIT) | |
| and the Vallado reference. Reimplemented Keplerian core; full drag/J2 | |
| secular terms are roadmap. Real ISS TLE parsed; second object synthetic. | |
| * CPA/TCPA conjunction screen (min-distance over propagated vectors). | |
| Our own min-distance math (same as the maritime CPA/TCPA organ). | |
| 5. RAVEN — authorize a workload at the disconnected edge (ROADMAP; substrate REAL) | |
| * TPM 2.0 PCR-extend + quote pattern (PCR[n] = SHA256(PCR[n-1] || measure)). | |
| Pattern reference: Keylime (Apache-2.0) remote-attestation model. Here | |
| the quote is computed and verified in-image; real TPM 2.0 hardware is a | |
| field step (labelled roadmap). | |
| * cosign offline image verification + conjunctive admission. Pattern | |
| reference: sigstore/cosign, UDS Core (Apache-2.0). | |
| SHARED PRIMITIVES (our own code) | |
| -------------------------------------------------------------------- | |
| * _KhipuChain: append-only SHA-256 chain H_n = SHA256(H_{n-1} || leaf_n), | |
| Merkle root (RFC-6962 leaf/node hashing pattern), inclusion proof + | |
| verification, and a single-byte tamper test that flips one character in | |
| one receipt field and proves the chain break + root mismatch + invalid | |
| inclusion proof. | |
| * _Timeline: real wall-clock per-step timing via time.perf_counter. | |
| PROVEN-FORMULA HONESTY (HARD DOCTRINE) | |
| -------------------------------------------------------------------- | |
| * Conjunctive GATE soundness = P2 (with CS1): PROVEN. | |
| * Tamper-evidence (append-only signed chain) = P5: PROVEN (gated on | |
| hashFn_collision_resistant). | |
| * Conformal interval coverage = W5-3 + W7-4: PROVEN; distribution-free; | |
| NEVER claims 100% coverage. | |
| * Trust score (Lambda) uniqueness = Conjecture 1: conditional / CI-green | |
| in strengthened classes; UNCONDITIONAL claim is FALSE. Not used as a | |
| pass/fail oracle here. | |
| * SLSA: L1 honest (signed provenance exists); L2 = roadmap, NOT claimed. | |
| DSSE signing uses a11oy's in-image ECDSA-P256-SHA256 key, verifiable against | |
| /cosign.pub. CANNONICO is a real, working mechanism today. The other four | |
| demos run the proven horizontal substrate on clearly labelled sample/replay | |
| data; the operational vertical (real registry, real feed, real TPM hardware, | |
| production ATO) is a fast stand-up and is labelled ROADMAP — never claimed done. | |
| Signed-off-by: Stephen P. Lutar Jr. <stephenlutar2@gmail.com> | |
| Co-Authored-By: Perplexity Computer Agent <agent@perplexity.ai> | |