fix(zdr): cite NVIDIA's own Trial ToS for the nvidia_nim not-ZDR classification

.github/workflows/opencode-review-dispatch.yml

@@ -5605,12 +5605,12 @@ jobs:
             }
 
             emit_known_missing_string_finding \
-              "STRIX_MODEL: contextual-orchestrator/orchestrator/auto" \
-              "Strix PR scans must use contextual-orchestrator/orchestrator/auto" \
+              "STRIX_MODEL: contextual-orchestrator/orchestrator/free" \
+              "Strix PR scans must use contextual-orchestrator/orchestrator/free" \
               ".github/workflows/strix.yml" \
               "scripts/ci/strix_required_workflow_smoke.sh"
             emit_known_missing_string_finding \
-              "Strix model overrides are limited to contextual-orchestrator/orchestrator/auto." \
+              "Strix model overrides are limited to contextual-orchestrator/orchestrator/free." \
               "Strix model overrides must remain gateway-only" \
               ".github/workflows/strix.yml" \
               "scripts/ci/strix_required_workflow_smoke.sh"

.github/workflows/strix.yml

@@ -549,14 +549,14 @@ jobs:
       - name: Gate Strix secrets
         id: gate
         env:
-          STRIX_MODEL: contextual-orchestrator/orchestrator/auto
+          STRIX_MODEL: contextual-orchestrator/orchestrator/free
           STRIX_MODEL_REQUESTED: ${{ github.event.client_payload.strix_llm || '' }}
         run: |
           requested_model="$(printf '%s' "$STRIX_MODEL_REQUESTED" | sed 's/^[[:space:]]*//;s/[[:space:]]*$//')"
           case "$requested_model" in
-            ""|orchestrator/auto|contextual-orchestrator/orchestrator/auto) ;;
+            ""|orchestrator/free|contextual-orchestrator/orchestrator/free) ;;
             *)
-              echo '::error::Strix model overrides are limited to contextual-orchestrator/orchestrator/auto.'
+              echo '::error::Strix model overrides are limited to contextual-orchestrator/orchestrator/free.'
               exit 1
               ;;
           esac
@@ -574,7 +574,7 @@ jobs:
           OPENROUTER_API_KEY: ${{ secrets.OPENROUTER_API_KEY }}
           OPENAI_API_KEY: ${{ secrets.OPENAI_API_KEY }}
           CONTEXTUAL_ORCHESTRATOR_REQUIRE_ZDR: ${{ steps.target_visibility.outputs.is_private }}
-          CONTEXTUAL_ORCHESTRATOR_POOL: auto
+          CONTEXTUAL_ORCHESTRATOR_POOL: free
         run: |
           set -euo pipefail
           bash "$TRUSTED_STRIX_SOURCE/scripts/ci/contextual_orchestrator_review_sidecar.sh"
@@ -713,11 +713,11 @@ jobs:
           strix_llm_file="$RUNNER_TEMP/strix_llm.txt"
           strix_model="$(printf '%s' "$STRIX_MODEL" | sed 's/^[[:space:]]*//;s/[[:space:]]*$//')"
           case "$strix_model" in
-            orchestrator/auto | contextual-orchestrator/orchestrator/auto)
-              printf '%s' 'orchestrator/auto' > "$strix_llm_file"
+            orchestrator/free | contextual-orchestrator/orchestrator/free)
+              printf '%s' 'orchestrator/free' > "$strix_llm_file"
               ;;
             *)
-              echo '::error::STRIX_LLM must select contextual-orchestrator/orchestrator/auto.'
+              echo '::error::STRIX_LLM must select contextual-orchestrator/orchestrator/free.'
               exit 1
               ;;
           esac

AGENTS.md

@@ -21,11 +21,12 @@ sidecar (`scripts/ci/contextual_orchestrator_review_sidecar.sh`). The five
 provider secrets (`BYTEZ_API_KEY`, `NVIDIA_NIM_API_KEY`,
 `NVIDIA_NIM_API_KEY_SUB`, `OPENROUTER_API_KEY`, `OPENAI_API_KEY`) enter its KV
 as bootstrap transport in the same process that discovers models and serves;
-OpenCode and Noema use the fail-closed zero-cost pool `orchestrator/free`.
-Authoritative Strix analysis uses the correctness-first `orchestrator/auto`
-provider-diverse pool. Non-free Strix routes
-are admitted only with complete published prompt/completion price and currency
-evidence, and private targets still require ZDR-compliant routes under
+OpenCode, Noema, and Strix all use the fail-closed zero-cost pool
+`orchestrator/free`. Strix uses the zero-cost `orchestrator/free` pool by
+explicit 2026-08-30 owner decision, superseding the prior `orchestrator/auto`
+(provider-diverse, non-free-admitting) default; private targets still require
+ZDR-compliant routes under
 [`scripts/ci/zdr_policy.py`](scripts/ci/zdr_policy.py).
-See [`docs/adr/0003-contextual-orchestrator-vendored-free-zdr.md`](docs/adr/0003-contextual-orchestrator-vendored-free-zdr.md).
+See [`docs/adr/0003-contextual-orchestrator-vendored-free-zdr.md`](docs/adr/0003-contextual-orchestrator-vendored-free-zdr.md)
+and its 2026-08-30 amendment.
 The materialization contract is also covered by [`docs/doctoring/exact-artifact-sbom-attestation.md`](docs/doctoring/exact-artifact-sbom-attestation.md).

CHANGELOG.md

@@ -5,6 +5,40 @@ this file. The format follows Keep a Changelog, and versioned releases follow
 Semantic Versioning where the repository publishes a release.
 
 ## [Unreleased]
+- Raise `contextual_orchestrator_review_sidecar.sh`'s
+  `ORCHESTRATOR_CATALOG_FAMILY_CAP` default from 4 to 8: root-caused the
+  live "no provider route passed the Strix plain-chat preflight" outage
+  blocking `noema-review`/`opencode-review`/`strix` org-wide to
+  `contextual_orchestrator_review_policy.py`'s family-cap candidate
+  selection deterministically admitting the same 4 alphabetically-first
+  `nvidia_nim`/`nvidia_nim_sub` free-model candidates on every run — 2 of
+  which are confirmed NVIDIA-retired model ids returning HTTP 404 forever —
+  while ~19 other healthy free candidates in the same discovery report
+  never got a chance. See the 2026-08-30 sidecar-preflight gap-baseline
+  entry for the full evidence trail, the exact trade-off reasoned through
+  (not live-verified, since this session lacks provider credentials), and
+  the more complete fix if this proves insufficient.
+- Switch Strix from `orchestrator/auto` to `orchestrator/free`, matching
+  OpenCode and Noema: `strix.yml`'s `STRIX_MODEL`/`CONTEXTUAL_ORCHESTRATOR_POOL`
+  default and both model-override allowlists, and
+  `scripts/ci/strix_quick_gate.sh`'s `is_contextual_orchestrator_model`, now
+  accept only `orchestrator/free`. This is an explicit, informed owner
+  override of `docs/adr/0003-contextual-orchestrator-vendored-free-zdr.md`'s
+  original `orchestrator/auto` decision (see that ADR's 2026-08-30
+  amendment and the matching gap-baseline entry for the full trade-off and
+  evidence trail): Strix no longer has a paid-model fallback and can go
+  fully dark during the class of single-provider-family-collapse incident
+  the original decision was written to survive, until the free-catalog's
+  stale-model and provider-diversity gaps are separately closed.
+- Strengthen `scripts/ci/zdr_policy.py`'s `nvidia_nim`/`nvidia_nim_sub` ZDR
+  attestation with a direct primary-source citation: NVIDIA's own current
+  *NVIDIA API Trial Terms of Service* (v. September 19, 2025), Section
+  3.3(iv), states User Content and Generated Content are collected "to
+  improve NVIDIA products and services, including AI models" — affirmative
+  evidence against zero data retention, not just an absence of attestation.
+  `zero_data_retention` stays `False` as it already was; only the citation
+  and note change. See the 2026-08-30 ZDR/NIM-routing gap-baseline entry for
+  the full architecture review this citation was part of.
 - Bump the vendored `contextual-orchestrator` review-sidecar pin from
   `5f2753a` (the #1422 pin) to current `main` `30c6d716`, picking up
   `ContextualWisdomLab/contextual-orchestrator#919`: generalizes the

docs/adr/0003-contextual-orchestrator-vendored-free-zdr.md

@@ -1,9 +1,11 @@
 # ADR-0003: Vendored contextual-orchestrator review sidecar with governed gateway pools
 
-- Status: accepted
+- Status: accepted, amended 2026-08-30 (see "2026-08-30 amendment" below — Strix
+  now uses `orchestrator/free`, not the `orchestrator/auto` this header
+  originally recorded)
 - Date: 2026-08-27
 - Scope: ContextualWisdomLab/.github central review pipelines (OpenCode autofix/dispatch + shared `opencode.jsonc` default + required Noema + Strix review)
-- Decision: Route every central CI review write/model execution that touches contracts in this repository through the **vendored** `contextual-orchestrator` gateway, served as a per-runner sidecar. OpenCode and Noema retain the fail-closed zero-cost virtual model id `orchestrator/free`; authoritative Strix security analysis uses the provider-diverse `orchestrator/auto` pool. Strix is intentionally correctness-first rather than zero-cost. **Zero Data Retention (ZDR)-compliant routes remain mandatory for private targets.**
+- Decision: Route every central CI review write/model execution that touches contracts in this repository through the **vendored** `contextual-orchestrator` gateway, served as a per-runner sidecar. OpenCode, Noema, and (as of the 2026-08-30 amendment) Strix all use the fail-closed zero-cost virtual model id `orchestrator/free`. **Zero Data Retention (ZDR)-compliant routes remain mandatory for private targets.**
 - Ownership: `.github` owns control-plane evidence; `ContextualWisdomLab/contextual-orchestrator` owns the gateway. The 2026-08-18 org decision (recorded in `ContextualWisdomLab/contextual-orchestrator` AGENTS.md) already migrated OpenCode/Noema/Strix to the orchestrator backend; this ADR is the org-repo (provider-config) half of that decision.
 - Figma File ID: N/A (no customer UI).
 
@@ -142,3 +144,35 @@ all five, and auto-optimize routing by cost.
   set `CONTEXTUAL_ORCHESTRATOR_REQUIRE_ZDR=true`; the catalog then excludes
   every non-ZDR route and fails closed when no attested ZDR route exists in the
   selected workflow pool.
+
+- **2026-08-30 amendment: Strix uses `orchestrator/free`, superseding this
+  ADR's original `orchestrator/auto` decision.** The org owner explicitly
+  directed Strix off the paid-inclusive `orchestrator/auto` pool and onto the
+  same zero-cost `orchestrator/free` pool OpenCode and Noema already use, so
+  no central review path executes a paid model. This is a deliberate,
+  informed override of the original decision above, not an oversight of it:
+  the trade-off the original decision recorded — "the 2026-08-29 exact-head
+  DiskSage scan proved that four discovered free routes all shared the
+  OpenRouter outage domain, which the gateway correctly collapsed to one
+  provider attempt... Strix has no external fallback" — was surfaced to the
+  owner explicitly, including a live 2026-08-30 reproduction of that same
+  single-family-collapse pattern (a `strix` run's `orchestrator/auto`
+  primary/free stage rejected 4/4 candidates — 2 timeouts, 2 HTTP 404s from
+  retired NVIDIA-hosted models — and only the `auto` pool's paid fallback
+  kept that run alive; see `docs/product-technical-gap-baseline.md`'s
+  2026-08-30 sidecar-preflight entries for the full evidence trail). The
+  owner's response, verbatim in substance: implement the free-only directive
+  as originally instructed. **Accepted consequence**: Strix has no external
+  fallback and can go fully dark (rather than degraded-but-running) during
+  the exact class of incident this ADR originally used `orchestrator/auto`
+  to survive, until the free-catalog's stale-model and provider-diversity
+  gaps documented alongside this amendment are separately closed. This is
+  the owner's accepted risk, not an unnoticed regression.
+  `scripts/ci/strix_quick_gate.sh`'s `is_contextual_orchestrator_model` no
+  longer accepts `orchestrator/auto`; `strix.yml`'s `STRIX_MODEL`/
+  `CONTEXTUAL_ORCHESTRATOR_POOL` default to `orchestrator/free`; and
+  `scripts/ci/strix_required_workflow_smoke.sh`/`AGENTS.md` were updated to
+  match. The `orchestrator/auto` pool mode itself is unchanged and still
+  exists in `contextual_orchestrator_review_policy.py`/the sidecar for any
+  other caller that opts into it explicitly — this amendment only removes it
+  as Strix's default and as an accepted Strix override value.