superseded by #1263: make Azure and cross-provider fallbacks executable

.github/workflows/strix.yml

@@ -696,6 +696,22 @@ jobs:
           printf '%s' 'https://models.github.ai/inference' > "$github_models_api_base_file"
           echo "STRIX_GITHUB_MODELS_API_BASE_FILE=$github_models_api_base_file" >> "$GITHUB_ENV"
 
+      - name: Prepare direct OpenAI fallback credentials
+        if: steps.gate.outputs.provider_mode == 'github_models' || steps.gate.outputs.provider_mode == 'openrouter' || steps.gate.outputs.provider_mode == 'nvidia_nim'
+        env:
+          OPENAI_FALLBACK_API_KEY: ${{ secrets.STRIX_OPENAI_API_KEY || secrets.OPENAI_API_KEY }}
+        run: |
+          umask 077
+          sanitized="$(printf '%s' "$OPENAI_FALLBACK_API_KEY" | tr -d '\r\n')"
+          trimmed="$(printf '%s' "$sanitized" | sed 's/^[[:space:]]*//;s/[[:space:]]*$//')"
+          if [ -z "$trimmed" ]; then
+            echo '::notice::No direct OpenAI key available; cross-provider Strix scans run without that fallback.'
+            exit 0
+          fi
+          openai_fallback_key_file="$RUNNER_TEMP/openai_fallback_key.txt"
+          printf '%s' "$trimmed" > "$openai_fallback_key_file"
+          echo "STRIX_OPENAI_FALLBACK_KEY_FILE=$openai_fallback_key_file" >> "$GITHUB_ENV"
+
       - name: Prepare Vertex AI credentials
         if: steps.gate.outputs.provider_mode == 'vertex_ai'
         env:
@@ -825,6 +841,7 @@ jobs:
           STRIX_FALLBACK_MODELS: ${{ steps.gate.outputs.provider_mode == 'github_models' && 'openai-direct/gpt-5.6-luna' || steps.gate.outputs.provider_mode == 'openai_direct' && 'openai-direct/gpt-5.6-luna' || steps.gate.outputs.provider_mode == 'openrouter' && 'openai-direct/gpt-5.6-luna' || steps.gate.outputs.provider_mode == 'nvidia_nim' && 'nvidia_nim/nvidia/llama-3.3-nemotron-super-49b-v1.5 openai-direct/gpt-5.6-luna' || '' }}
           STRIX_GITHUB_MODELS_API_BASE_FILE: ${{ env.STRIX_GITHUB_MODELS_API_BASE_FILE }}
           STRIX_GITHUB_MODELS_KEY_FILE: ${{ env.STRIX_GITHUB_MODELS_KEY_FILE }}
+          STRIX_OPENAI_FALLBACK_KEY_FILE: ${{ env.STRIX_OPENAI_FALLBACK_KEY_FILE }}
           STRIX_FAIL_ON_PROVIDER_SIGNAL: "1"
           STRIX_VERTEX_FALLBACK_MODELS: ""
           NPM_CONFIG_IGNORE_SCRIPTS: "true"

CHANGELOG.md

@@ -37,6 +37,16 @@ Semantic Versioning where the repository publishes a release.
 
 ### Changed
 
+- Route the exact single-line LiteLLM/Azure unsupported-temperature failure to
+  an already-configured distinct Strix fallback without accepting split-line
+  imitations, and normalize the workflow's `openai-direct/` fallback alias to
+  LiteLLM's `openai_direct/` provider prefix before dispatch. Cross-provider
+  attempts now switch to the trusted OpenAI credential and clear the primary
+  provider API base, without turning an incomplete provider scan into passing
+  evidence. Exact clean model-quality and Hugging Face advisories are filtered
+  consistently from console and report logs, while any appended warning text
+  remains fail closed. The incident and fail-closed boundary are recorded in
+  `docs/doctoring/strix-unsupported-sampling-fallback.md`.
 - Emit completed repository pull-list requests as they finish in the five-minute
   agent-mention sweep, while retaining the four-worker ceiling, rotation, and
   exact-name dispatch ledger, so one slow repository cannot hide ready sibling

docs/doctoring/strix-nvidia-nim-not-found-fallback.md

@@ -30,6 +30,12 @@ combining with an unrelated application `404` to spoof infrastructure fallback.
 Provider-side failure also remains a fail-closed incomplete scan until a distinct
 fallback produces complete evidence.
 
+A completed zero-vulnerability fallback may emit Strix's model-quality banner
+and Hugging Face's exact unauthenticated-download advisory while loading public
+scanner assets. The trusted gate removes only those two complete lines from both
+console and structured report logs. A suffix, prefix, or any other warning text
+remains fail-closed provider-failure evidence.
+
 Exhausted provider infrastructure remains fail-closed even when the trusted
 gate has classified every observed threshold finding as outside the pull
 request's changed files. That classification scopes authoritative findings; it
@@ -53,7 +59,11 @@ Regression evidence proves that:
 8. provider exhaustion remains non-passing after unchanged baseline findings;
 9. changed, unmapped, and changed-manifest findings also block after provider
    exhaustion; and
-10. the required-workflow smoke contract pins these properties.
+10. the exact model-quality and Hugging Face download advisories do not reject
+    an otherwise complete zero-vulnerability fallback when they appear in
+    console or report logs;
+11. appended warning text is not hidden by either exact advisory filter; and
+12. the required-workflow smoke contract pins these properties.
 
 ## Limitations
 

docs/doctoring/strix-unsupported-sampling-fallback.md

@@ -0,0 +1,67 @@
+# Strix unsupported sampling-parameter fallback
+
+## Observed failure
+
+An Azure `gpt-5.6-sol` Strix run failed before vulnerability analysis because
+LiteLLM sent `temperature=0.2`. Azure accepts only the model default of `1`,
+and LiteLLM had no fallback group for the selected model. Microsoft documents
+`temperature` as unsupported for GPT-5 reasoning models (Microsoft, 2026),
+while the pinned Strix configuration surface exposes no generation-parameter
+control (usestrix, n.d.).
+
+## Root cause and repair
+
+The preferred request-boundary repair is to omit a sampling parameter that a
+caller did not explicitly provide. `contextual-orchestrator` owns that provider
+boundary for organization software. The pinned Strix integration cannot yet
+express the omission through its documented configuration, so the trusted
+quick gate recognizes only one physical error line containing all of these
+signals:
+
+- a LiteLLM `BadRequestError`;
+- Azure or OpenAI exception context;
+- the unsupported `temperature` value and supported default; and
+- the missing LiteLLM fallback model group.
+
+That exact capability failure is infrastructure evidence and may move directly
+to an already-configured distinct outer fallback. It is not eligible for a
+same-model retry. The shared model normalizer also translates the workflow's
+human-readable `openai-direct/` selector into LiteLLM's `openai_direct/`
+provider prefix before dispatch. A cross-provider direct OpenAI fallback reads
+the established OpenAI secret from a trusted runtime file and clears the
+primary provider's API base; otherwise a NVIDIA or OpenRouter run would send
+the fallback to the wrong endpoint with the wrong credential. If the fallback
+credential is unavailable, the attempted fallback fails configuration closed.
+If no distinct fallback exists or every fallback fails, the required Strix
+check remains non-passing. Existing changed, unmapped, manifest,
+`ModelBehaviorError`, and vulnerability-report boundaries remain fail closed.
+
+Cross-line signal assembly is deliberately rejected so unrelated target output
+cannot manufacture a provider capability error from separate log lines.
+
+## Verification
+
+- The reproduced single-line Azure failure reaches the configured GitHub
+  Models fallback exactly once and succeeds only when that scan completes.
+- The configured `openai-direct/gpt-5.6-luna` fallback normalizes to the
+  LiteLLM-compatible `openai_direct/gpt-5.6-luna` selector.
+- A NVIDIA-primary run dispatches that fallback with the OpenAI credential and
+  no inherited NVIDIA API base.
+- A split-line imitation is non-recoverable and never dispatches the fallback.
+- The full Python suite, native workflow validation, Bash syntax checks, and
+  complete Strix shell regression suite run on the final tree.
+
+## References
+
+AkikoOrenji. (2026, June 4). *[Feature] Expose LLM generation parameters to
+control local/OpenAI-compatible model behaviour and prevent runaway tool-call
+loops* (Issue No. 514) [GitHub issue]. GitHub.
+https://github.com/usestrix/strix/issues/514
+
+Microsoft. (2026, August 20). *Azure OpenAI reasoning models—GPT-5 series,
+o3-mini, o1, o1-mini*. Microsoft Learn.
+https://learn.microsoft.com/en-us/azure/foundry/openai/how-to/reasoning
+
+usestrix. (n.d.). *Configuration* [Computer software documentation]. GitHub.
+Retrieved August 23, 2026, from
+https://github.com/usestrix/strix/blob/main/docs/advanced/configuration.mdx

scripts/ci/strix_model_utils.sh

@@ -105,6 +105,10 @@ normalize_model() {
 		printf '%s\n' "$model"
 		return 0
 		;;
+	openai-direct/?*)
+		printf 'openai_direct/%s\n' "${model#openai-direct/}"
+		return 0
+		;;
 	*/*)
 		printf '%s\n' "$model"
 		return 0

scripts/ci/strix_quick_gate.sh

@@ -172,6 +172,16 @@ known_internal_warning = re.compile(
     r"|ended a turn without a lifecycle tool call \(interactive=False\)"
     r"); forcing tool continuation \(\d+/\d+\): "
 )
+known_clean_advisory = re.compile(
+    r"^(?:[ \t│]*MODEL QUALITY WARNING[ \t│]*"
+    r"|Warning: You are sending unauthenticated requests to the HF Hub\. "
+    r"Please set a HF_TOKEN to enable higher rate limits and faster downloads\.)$"
+)
+known_optional_web_search_advisory = re.compile(
+    r"^\d{4}-\d{2}-\d{2} \d{2}:\d{2}:\d{2}\.\d+ WARNING "
+    r"[^ ]+ - strix\.tools\.web_search\.tool: "
+    r"web_search invoked without PERPLEXITY_API_KEY configured$"
+)
 
 
 def iter_report_logs(root: Path):
@@ -194,7 +204,13 @@ for log_path in iter_report_logs(root):
         lines = log_path.read_text(encoding="utf-8").splitlines(keepends=True)
     except UnicodeDecodeError:
         continue
-    filtered = [line for line in lines if not known_internal_warning.match(line)]
+    filtered = [
+        line
+        for line in lines
+        if not known_internal_warning.match(line)
+        and not known_clean_advisory.fullmatch(line.rstrip("\r\n"))
+        and not known_optional_web_search_advisory.fullmatch(line.rstrip("\r\n"))
+    ]
     if filtered != lines:
         log_path.write_text("".join(filtered), encoding="utf-8")
 PY
@@ -380,6 +396,23 @@ if [ -n "$STRIX_GITHUB_MODELS_KEY_FILE" ]; then
 	fi
 fi
 
+STRIX_OPENAI_FALLBACK_KEY_FILE="${STRIX_OPENAI_FALLBACK_KEY_FILE:-}"
+if [ -n "$STRIX_OPENAI_FALLBACK_KEY_FILE" ] && { [ ! -f "$STRIX_OPENAI_FALLBACK_KEY_FILE" ] || [ -L "$STRIX_OPENAI_FALLBACK_KEY_FILE" ]; }; then
+	echo "ERROR: STRIX_OPENAI_FALLBACK_KEY_FILE must reference a regular file containing the API key." >&2
+	exit 2
+fi
+if [ -n "$STRIX_OPENAI_FALLBACK_KEY_FILE" ] && ! STRIX_OPENAI_FALLBACK_KEY_FILE="$(resolve_trusted_input_file "STRIX_OPENAI_FALLBACK_KEY_FILE" "$STRIX_OPENAI_FALLBACK_KEY_FILE")"; then
+	exit 2
+fi
+STRIX_OPENAI_FALLBACK_KEY=""
+if [ -n "$STRIX_OPENAI_FALLBACK_KEY_FILE" ]; then
+	STRIX_OPENAI_FALLBACK_KEY="$(trim_whitespace "$(cat -- "$STRIX_OPENAI_FALLBACK_KEY_FILE")")"
+	if [ -z "$STRIX_OPENAI_FALLBACK_KEY" ]; then
+		echo "ERROR: STRIX_OPENAI_FALLBACK_KEY_FILE must contain a non-empty API key." >&2
+		exit 2
+	fi
+fi
+
 require_non_negative_integer() {
 	local value="$1"
 	local label="$2"
@@ -2380,6 +2413,12 @@ resolved_llm_api_base_for_model() {
 	if is_vertex_model "$model"; then
 		return 0
 	fi
+	case "$(normalize_model "$model"):$PRIMARY_MODEL" in
+	openai_direct/*:openai_direct/*) ;;
+	openai_direct/*:*)
+		return 0
+		;;
+	esac
 
 	local api_base_file="$LLM_API_BASE_FILE"
 	local api_base_file_name="LLM_API_BASE_FILE"
@@ -2490,20 +2529,32 @@ run_strix_once() {
 	if ! llm_api_base_value="$(resolved_llm_api_base_for_model "$model")"; then
 		return 2
 	fi
-	child_model="$(child_model_for_api_base "$model" "$llm_api_base_value")"
+	local normalized_model
+	normalized_model="$(normalize_model "$model")"
+	child_model="$(child_model_for_api_base "$normalized_model" "$llm_api_base_value")"
 	if ! resolved_target_path="$(resolve_current_target_path "$TARGET_PATH")"; then
 		return 1
 	fi
 	local start_epoch
 	start_epoch="$(date +%s)"
 	local child_llm_api_key=""
-	if ! is_vertex_model "$(normalize_model "$model")"; then
+	if ! is_vertex_model "$normalized_model"; then
 		child_llm_api_key="$LLM_API_KEY"
-		if is_github_models_model "$(normalize_model "$model")" && [ -n "$STRIX_GITHUB_MODELS_KEY" ]; then
+		if is_github_models_model "$normalized_model" && [ -n "$STRIX_GITHUB_MODELS_KEY" ]; then
 			# Cross-provider fallback: github_models/* models authenticate
 			# with the GitHub Models token, not the direct-OpenAI key.
 			child_llm_api_key="$STRIX_GITHUB_MODELS_KEY"
 		fi
+		case "$normalized_model:$PRIMARY_MODEL" in
+		openai_direct/*:openai_direct/*) ;;
+		openai_direct/*:*)
+			if [ -z "$STRIX_OPENAI_FALLBACK_KEY" ]; then
+				echo "ERROR: direct OpenAI fallback requires STRIX_OPENAI_FALLBACK_KEY_FILE." >&2
+				return 2
+			fi
+			child_llm_api_key="$STRIX_OPENAI_FALLBACK_KEY"
+			;;
+		esac
 	fi
 	set -o pipefail
 	set +e
@@ -2821,6 +2872,19 @@ is_nvidia_nim_not_found_error() {
 	return 1
 }
 
+is_unsupported_model_parameter_error() {
+	# Strix currently has no generation-parameter override. Match the exact
+	# single-line LiteLLM/Azure capability failure so a reasoning model that
+	# rejects Strix's temperature can move to the already-configured fallback.
+	if grep -Ei 'litellm(\.exceptions)?\.BadRequestError' "$STRIX_LOG" |
+		grep -Ei '(AzureException|OpenAIException)' |
+		grep -Eiq "Unsupported value:[[:space:]]*['\"]temperature['\"].*Only the default[[:space:]]*\\(1\\)[[:space:]]*value is supported.*No fallback model group found"; then
+		return 0
+	fi
+
+	return 1
+}
+
 is_model_behavior_error() {
 	# Classify only a module-qualified Strix/Agents SDK protocol exception.
 	# A bare source-file mention of ModelBehaviorError is not retryable.
@@ -3151,7 +3215,8 @@ is_llm_token_limit_error() {
 # was interrupted or incomplete.  Used as a guard to prevent the
 # below-threshold override from silently passing an aborted scan.
 has_detected_infrastructure_error() {
-	if grep -Eiq '(^|[^[:alpha:]])(Fatal|Denied|Warn|Warning)([^[:alpha:]]|$)' "$STRIX_LOG"; then
+	if grep -Eiq '(^|[^[:alpha:]])(Fatal|Denied|Warn|Warning)([^[:alpha:]]|$)' \
+		< <(grep -Eiv '^[[:space:]│]*MODEL QUALITY WARNING[[:space:]│]*$|^Warning: You are sending unauthenticated requests to the HF Hub\. Please set a HF_TOKEN to enable higher rate limits and faster downloads\.$' "$STRIX_LOG"); then
 		return 0
 	fi
 
@@ -3183,6 +3248,10 @@ has_detected_infrastructure_error() {
 		return 0
 	fi
 
+	if is_unsupported_model_parameter_error; then
+		return 0
+	fi
+
 	if is_model_behavior_error; then
 		return 0
 	fi
@@ -4076,6 +4145,10 @@ is_model_retryable_error() {
 		return 0
 	fi
 
+	if is_unsupported_model_parameter_error; then
+		return 0
+	fi
+
 	# A provider failure can be recorded only in Strix's structured report log.
 	# run_strix_once already marks that evidence as infrastructure failure, but
 	# the child stdout log used by the classifiers may not contain the provider