10 concrete patterns (Directional Collapse, Constant Class, Corpus Uniformity
Trap, Capacity Plateau, Schema Drift Contamination, Upstream Feature Starvation,
Axis Factorization, Grokking Phase Transition, Guardrail Rescue, Label Boundary
Drift), each with observable signal, mechanism, detection test, fix, failure
conditions, and a real whetstone_DSL example where one exists.
Motivated by the automatability gate session (April 2026) where Directional
Collapse was the root cause — deterministic always confused with template 100%
directionally, resolved by 2-stage decomposition to 100% accuracy.
Intended for use by developers and model-assisted debuggers without deep ML
background. Scale is not a pattern; these patterns exist to refute it.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- split_prereq_op_dataset.py: extract accepted/rejected/schema-drift rows per gate
- gen_prereq_op_rsa_data.py: convert accepted rows to Fabricate TSV format
- 1000-run extracted dataset: 1297 accepted prereq_op rows, 24 schema-drift
- Key finding: needs_validate_intake always True; effective gate is binary
(needs_resolve_dependencies 77%, needs_architect_review 20%)
- Key finding: run corpus text is template-driven (17 unique inputs) — not
suitable for text-based training; RSA exists before ideal input is available
- ROADMAP: gate lifecycle framing — goal is choice of backend, not forced
determinism; RSA operates on imperfect/mixed/incomplete input by design
- training_data_strategy.md: taskitem-as-self-contained-work-unit constraint
- SPRINT-003: updated with 1000-run results, text audit findings, design constraints
- SPRINT-004: pipeline decision audit — manual walkthrough of every decision
from project description to generated artifact; execute next session
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
