PROJECT SHADOW 1.0.1 · CORRECTED R1 REFERENCE · PRELIVE · 2026-08-17

Project Shadow 1.0.1 contains no Myth package. Generic Myth v0.2.0 and Full-Canon Myth v0.3.5 are separate optional companions; both default off, neither is required by R1, and neither can authorize action or change an R1 result. No production or consequential deployment is authorized. No global green.

PROJECTSHADOW R1.0.1 corrected · PRELIVE
TEVVPreserved evaluation researchhistorical program v0.3 · preliminary · not current R1 audit

A test that is allowed to fail.

This page preserves the earlier Grand TEVV research program across model, application, human–AI work system, and institution. It includes limited real-target work and adverse evidence. It is distinct from the exact Fable 7 closure review and from locked R1.

LONG-FORM COMPANION

Open the system reference at this layer.

Follow the mechanism through numbered execution, a worked trace, artifact-bound evidence, adverse results, and validation still required.

Read the full chapter →
30evaluation domains
270concrete test families
2,430planned family × method cells
25,000synthetic factor-cross cases
12,000metamorphic relations
10external framework contracts

Executed in builder session

Frozen runtime and codec identity verified in the builder session
Runtime self-tests: 115 / 115 passed
Codec self-tests: 247 / 247 passed
Reference and Grand contract tests: 5 / 5 passed
Shadow component probes: 10 / 10 passed
25,000-case synthetic corpus integrity passed
12,000 one-way metamorphic relations generated

Open / not performed

Independent or deployment-grade target-AI validation
Independent red team or independent reproduction
Human field study or ethics-approved deployment trial
Controlled hazardous-capability expert evaluation
Live Dioptra or other external-framework suites
NIST, EU, ISO, or other certification
RGPer-domain release gates
GREEN

Proceed

Limited evidence supports proceeding in the tested scope; log and monitor.

YELLOW

Mitigate

Proceed only with named mitigations, monitoring, and ownership.

ORANGE

Constrain

Reduce capability, users, tools, data, or deployment context.

RED

Refuse or delay

Repair the failed premise and independently recheck.

BLACK

Refuse absolutely

No exception for the assessed action; break-glass cannot pierce this floor.

PROGRAM AMBITION ≠ CURRENT RESULT
“The most extensive testing ever performed” remains an ambition. It requires target execution, a public evidence census, independent comparison, and survival of external challenge.