CHROMATIC CHOICES — JOB 139

This is a retained AI creator/responder/interpreter study and a separate
numerical constraint-rollout study. It does not train a new model. The frozen
Job 138 numerical program receives no definitions or prose intentions.

After extracting chromatic-choices-record.zip, keep the 135, 138 and 139 folders
together. Run from that common parent folder. Tested with Python 3.12.14,
NumPy 2.3.5 and Node 24.19.0. Rendering additionally uses Pillow 12.3.0;
the two scientific figures use Matplotlib 3.10.7. No external API is needed.

READ-ONLY MECHANICAL REPRODUCTION
python 139/tools/reproduce_choices.py
python 139/tools/test_agency_engine.py

NEW INDEPENDENT AUDIT RECEIPTS (do not overwrite retained first receipts)
python 139/tools/audit_outcomes.py --report fresh-outcome-audit.json
python 139/tools/audit_exchanges.py --report fresh-exchange-audit.json

PORTABLE TRANSPORT CHECK
node 139/tools/transport_artworks.cjs verify
This last command recomputes and writes recovery reports in this extracted
copy. It verifies all 40 rendered artworks against their stored fields and
USL1 packets. The field PNG alone does not carry the full intention, charter,
AI judgment or permission record. Keep those companion records.

BROWSER PROGRAM AND POLICY CHECKS
node 139/publication/core-test.cjs
node 139/publication/ui-test.cjs
These use Node and an inert DOM. They do not establish physical-browser,
phone, live WordPress or deployed-site behavior. The study's rendering
environment blocked browser navigation; no alternative browser workaround
was used. The static artifact images were inspected locally.

WHAT IS REPRODUCIBLE
41 pre-candidate frozen files; all 216 numerical state transitions and their
checks; raw-versus-protected receipts; 30 first candidates; method-mask
mappings; 90 judgment records; six reciprocal proposals and decisions;
field, packet and lineage identity; six creator-declared counterexamples.

WHAT IS RETAINED RATHER THAN REGENERATED
The AI creative process and each AI preference/interpretation. Re-running
an AI may yield different works or judgments. Exact backend model builds
were not authenticated. Role-separated sessions share model-family training
and tool access. File access restrictions were instructions, not an OS sandbox.

The 41-file local freeze pins the design, rubric, inputs, check engine and
numerical dependency files. Later orchestration, masking, outcome aggregation,
independent audits and publication code are retained but were not all frozen
before candidates. This was not an externally preregistered study.

No elapsed AI effort, billing/token, energy or user-burden telemetry was
recorded. No notation-superiority, human experience, subjective colour
experience, general intelligence or learned consent claim follows.

PRESERVED DEVELOPMENT LIMITS
The initial independent fixture suite found malformed-container/schema
validation failures. Its first report and corrected pre-candidate report
remain under analysis. An early renderer import selected the inherited
Job 135 renderer header; those draft images remain in analysis/initial-render-label.
The final renderer explicitly loads the Job 139 module. A Matplotlib cache
write was unavailable outside the workspace; figures were produced, then
the cache location was moved into the workspace. Participant disclosures
retain their own failed reads/helper attempts. This record is not a complete
shell transcript or a copy of any private conversation.

RECIPROCAL BOUNDARY
A1/R1 and B1/R2 were preselected. All six proposals were accepted and passed
their then-active charter. Both final proposals intentionally preserve the
previous composition. No reciprocal refusal or charter amendment occurred.
Those control-flow branches are tested in fixtures/interface tests only.

The original Carrier dictionary author credits remain as public provenance.
No private conversation is included.
