# Stage 136: 9->N Rubrics & System-1 Gate (aq136_evaluate_extensible_rubric_and_system1 / DB-UNIT-17)

Evaluates priority-ordered 9-to-N extensible rubrics with knockout short-circuiting and routes 4-rung System-1 CPU tasks gated by 2% blind canaries.

- Atomic micro-queue: `aq136_evaluate_extensible_rubric_and_system1` (visibility: `public_commons`)
- Backing database unit: `DB-UNIT-17` (`Cascading Decision Models, Local Classifiers & Abstention Core`)
- Database cell: `lsh-db.teleon.dev` (`teleon_core.decision_rubric_schemas`, `teleon_core.cascade_classification_ledger`, `teleon_core.abstention_calibration_sets`)
- Claim row hydration: `single_self_contained_flat_task_row` (100% context inline, normalized writeback)
- Track 2 Sister queue (`SISTER-17`): `sq17-improve-decisions-queue.teleon.dev` (Distill frontier decisions into compact SetFit/ModernBERT/GLiNER2 CPU heads and tighten natural no-match abstention thresholds.)
- Lanes: PrioritizedExtensibleList, Knockout Short-Circuit, 2% Canary Brier/ECE

## Execution engines & AI Agent freedom contract

- **Deterministic (0 model calls)**: `rule_model2vec_setfit_gliner_cpu_cascade_v1` (Gain/cost rubric sorter, Schmitt verdict gate & Brier/ECE canary checker (0 calls))
- **Hybrid (<=1 model call)**: `clef_jev_clairic_typed_decision_escalation_v1`
- **AI Agent & Bot Fleet (1..6 calls)**: `ambiguous_tail_reasoning_arbitrator_v1` (Domain-specific extension criterion & calibration holdout author)
  - Contract: **Instructions + Resources + Freedom to search for new resources and new ways not bound by instructions**

## Database invariants

- Static rubrics precede dynamic item state so input-only decision APIs capture cached-prefix discounts.
- Below-threshold confidence returns explicit ABSTAIN or escalates rather than pushing mismatched components.

## External & internal bot fleet actions

- Append custom domain criteria beyond the 9 defaults without altering caller schemas
- Short-circuit expensive downstream rubric checks the instant a knockout gate fails
- Audit client-worker CPU classification batches against 2% blind canaries (Brier <= 0.18, ECE <= 0.08)

## Sister self-tuning areas

- Distill recurring multi-question rubrics into 8-to-32-shot SetFit and ONNX INT8 CPU classifiers (DIM-03, DIM-06)
- Calibrate per-domain abstention thresholds on 25%+ natural no-suitable-component holdouts (DIM-01, DIM-06)
- Benchmark choice-order invariance across Clef-flash, Jev, Clairic, and Decisions endpoints (DIM-02, DIM-08)