feat: LLM dispatch wiring, eval models, bifrost config loader [OMN-6785]

src/omnibase_infra/enums/__init__.py

@@ -86,6 +86,10 @@
 from omnibase_infra.enums.enum_dedupe_strategy import EnumDedupeStrategy
 from omnibase_infra.enums.enum_dispatch_status import EnumDispatchStatus
 from omnibase_infra.enums.enum_environment import EnumEnvironment
+from omnibase_infra.enums.enum_eval_finding_category import EnumEvalFindingCategory
+from omnibase_infra.enums.enum_eval_finding_severity import EnumEvalFindingSeverity
+from omnibase_infra.enums.enum_eval_task_status import EnumEvalTaskStatus
+from omnibase_infra.enums.enum_eval_task_type import EnumEvalTaskType
 from omnibase_infra.enums.enum_execution_shape_violation import (
     EnumExecutionShapeViolation,
 )
@@ -151,6 +155,10 @@
     "EnumDedupeStrategy",
     "EnumDispatchStatus",
     "EnumEnvironment",
+    "EnumEvalFindingCategory",
+    "EnumEvalFindingSeverity",
+    "EnumEvalTaskStatus",
+    "EnumEvalTaskType",
     "EnumExecutionShapeViolation",
     "EnumHandlerErrorType",
     "EnumHandlerLoaderError",

src/omnibase_infra/enums/enum_eval_finding_category.py

@@ -0,0 +1,28 @@
+# SPDX-FileCopyrightText: 2025 OmniNode.ai Inc.
+# SPDX-License-Identifier: MIT
+"""Eval finding category enum.
+
+Related:
+    - OMN-6795: Define eval task models and enums
+
+.. versionadded:: 0.29.0
+"""
+
+from __future__ import annotations
+
+from enum import Enum
+
+
+class EnumEvalFindingCategory(str, Enum):
+    """Category of an evaluation finding."""
+
+    BUG = "bug"
+    TECH_DEBT = "tech_debt"
+    DOC_STALE = "doc_stale"
+    REGRESSION_RISK = "regression_risk"
+    SECURITY = "security"
+    PERFORMANCE = "performance"
+    STYLE = "style"
+
+
+__all__: list[str] = ["EnumEvalFindingCategory"]

src/omnibase_infra/enums/enum_eval_finding_severity.py

@@ -0,0 +1,25 @@
+# SPDX-FileCopyrightText: 2025 OmniNode.ai Inc.
+# SPDX-License-Identifier: MIT
+"""Eval finding severity enum.
+
+Related:
+    - OMN-6795: Define eval task models and enums
+
+.. versionadded:: 0.29.0
+"""
+
+from __future__ import annotations
+
+from enum import Enum
+
+
+class EnumEvalFindingSeverity(str, Enum):
+    """Severity level of an evaluation finding."""
+
+    HIGH = "high"
+    MEDIUM = "medium"
+    LOW = "low"
+    INFO = "info"
+
+
+__all__: list[str] = ["EnumEvalFindingSeverity"]

src/omnibase_infra/enums/enum_eval_task_status.py

@@ -0,0 +1,26 @@
+# SPDX-FileCopyrightText: 2025 OmniNode.ai Inc.
+# SPDX-License-Identifier: MIT
+"""Eval task status enum.
+
+Related:
+    - OMN-6795: Define eval task models and enums
+
+.. versionadded:: 0.29.0
+"""
+
+from __future__ import annotations
+
+from enum import Enum
+
+
+class EnumEvalTaskStatus(str, Enum):
+    """Status of an evaluation task."""
+
+    PENDING = "pending"
+    RUNNING = "running"
+    COMPLETED = "completed"
+    FAILED = "failed"
+    BUDGET_EXHAUSTED = "budget_exhausted"
+
+
+__all__: list[str] = ["EnumEvalTaskStatus"]

src/omnibase_infra/enums/enum_eval_task_type.py

@@ -0,0 +1,41 @@
+# SPDX-FileCopyrightText: 2025 OmniNode.ai Inc.
+# SPDX-License-Identifier: MIT
+"""Eval task type enum for autonomous off-peak evaluations.
+
+Defines the types of evaluation tasks that can be dispatched to
+cheap LLM providers (Gemini Flash, GLM, local small models) during
+off-peak periods.
+
+Related:
+    - OMN-6795: Define eval task models and enums
+
+.. versionadded:: 0.29.0
+"""
+
+from __future__ import annotations
+
+from enum import Enum
+
+
+class EnumEvalTaskType(str, Enum):
+    """Types of autonomous evaluation tasks.
+
+    Each type maps to a specific prompt template and evaluation
+    strategy in ``ServiceEvalRunner``.
+
+    Attributes:
+        CODE_REVIEW: Automated code review of recently changed files.
+        TECH_DEBT_SCAN: Scan for tech debt patterns and anti-patterns.
+        DOC_FRESHNESS: Check documentation freshness against code changes.
+        REGRESSION_TEST: Run regression test analysis on changed paths.
+        HOSTILE_REVIEW: Adversarial review looking for subtle bugs.
+    """
+
+    CODE_REVIEW = "code_review"
+    TECH_DEBT_SCAN = "tech_debt_scan"
+    DOC_FRESHNESS = "doc_freshness"
+    REGRESSION_TEST = "regression_test"
+    HOSTILE_REVIEW = "hostile_review"
+
+
+__all__: list[str] = ["EnumEvalTaskType"]

src/omnibase_infra/models/eval/__init__.py

@@ -1,21 +1,33 @@
 # SPDX-FileCopyrightText: 2025 OmniNode.ai Inc.
 # SPDX-License-Identifier: MIT
-"""Auto-eval models for autonomous LLM evaluation (OMN-6795).
+"""Eval models for autonomous evaluation tasks.
 
 Exports:
-    ModelAutoEvalTask: Defines a single evaluation task to run against an LLM endpoint.
-    ModelAutoEvalResult: Result of executing an eval task, with scores and cost.
-    ModelAutoEvalBudgetCap: Budget constraints for eval execution windows.
+    ModelAutoEvalTask: Defines a single auto-evaluation task to run against an LLM endpoint.
+    ModelAutoEvalResult: Result of executing an auto-eval task, with scores and cost.
+    ModelAutoEvalBudgetCap: Budget constraints for auto-eval execution windows.
+    ModelEvalBudgetCap: Budget cap for eval tasks.
+    ModelEvalFinding: Individual eval finding.
+    ModelEvalResult: Result of an eval run.
+    ModelEvalTask: Eval task definition.
 """
 
 from omnibase_infra.models.eval.model_auto_eval_budget_cap import (
     ModelAutoEvalBudgetCap,
 )
 from omnibase_infra.models.eval.model_auto_eval_result import ModelAutoEvalResult
 from omnibase_infra.models.eval.model_auto_eval_task import ModelAutoEvalTask
+from omnibase_infra.models.eval.model_eval_budget_cap import ModelEvalBudgetCap
+from omnibase_infra.models.eval.model_eval_finding import ModelEvalFinding
+from omnibase_infra.models.eval.model_eval_result import ModelEvalResult
+from omnibase_infra.models.eval.model_eval_task import ModelEvalTask
 
 __all__: list[str] = [
-    "ModelAutoEvalTask",
-    "ModelAutoEvalResult",
     "ModelAutoEvalBudgetCap",
+    "ModelAutoEvalResult",
+    "ModelAutoEvalTask",
+    "ModelEvalBudgetCap",
+    "ModelEvalFinding",
+    "ModelEvalResult",
+    "ModelEvalTask",
 ]

src/omnibase_infra/models/eval/model_eval_budget_cap.py

@@ -0,0 +1,47 @@
+# SPDX-FileCopyrightText: 2025 OmniNode.ai Inc.
+# SPDX-License-Identifier: MIT
+"""Budget cap model for autonomous off-peak evaluations.
+
+Related:
+    - OMN-6795: Define eval task models and enums
+
+.. versionadded:: 0.29.0
+"""
+
+from __future__ import annotations
+
+from pydantic import BaseModel, ConfigDict, Field
+
+
+class ModelEvalBudgetCap(BaseModel):
+    """Budget cap for autonomous evaluation tasks.
+
+    Controls spending across evaluation windows to prevent runaway costs.
+
+    Attributes:
+        max_tokens_per_window: Maximum total tokens across all tasks in a window.
+        max_cost_usd_per_window: Maximum total cost in USD per window.
+        window_hours: Duration of the budget window in hours.
+    """
+
+    model_config = ConfigDict(frozen=True, extra="forbid")
+
+    max_tokens_per_window: int = Field(
+        default=500_000,
+        ge=1000,
+        description="Max tokens per budget window.",
+    )
+    max_cost_usd_per_window: float = Field(
+        default=1.0,
+        ge=0.01,
+        description="Max cost in USD per budget window.",
+    )
+    window_hours: int = Field(
+        default=24,
+        ge=1,
+        le=168,
+        description="Budget window duration in hours.",
+    )
+
+
+__all__: list[str] = ["ModelEvalBudgetCap"]

src/omnibase_infra/models/eval/model_eval_finding.py

@@ -0,0 +1,41 @@
+# SPDX-FileCopyrightText: 2025 OmniNode.ai Inc.
+# SPDX-License-Identifier: MIT
+"""Eval finding model for autonomous off-peak evaluations.
+
+Related:
+    - OMN-6795: Define eval task models and enums
+
+.. versionadded:: 0.29.0
+"""
+
+from __future__ import annotations
+
+from pydantic import BaseModel, ConfigDict, Field
+
+from omnibase_infra.enums.enum_eval_finding_category import EnumEvalFindingCategory
+from omnibase_infra.enums.enum_eval_finding_severity import EnumEvalFindingSeverity
+
+
+class ModelEvalFinding(BaseModel):
+    """A single finding from an evaluation task.
+
+    Attributes:
+        severity: Finding severity level.
+        category: Finding category.
+        file_path: File path where the finding was detected.
+        line_number: Line number (0 if not applicable).
+        description: Human-readable description of the finding.
+        suggestion: Suggested fix or action.
+    """
+
+    model_config = ConfigDict(frozen=True, extra="forbid")
+
+    severity: EnumEvalFindingSeverity = Field(..., description="Severity level.")
+    category: EnumEvalFindingCategory = Field(..., description="Finding category.")
+    file_path: str = Field(default="", description="File path of the finding.")
+    line_number: int = Field(default=0, ge=0, description="Line number.")
+    description: str = Field(..., description="Description of the finding.")
+    suggestion: str = Field(default="", description="Suggested fix.")
+
+
+__all__: list[str] = ["ModelEvalFinding"]

src/omnibase_infra/models/eval/model_eval_result.py

@@ -0,0 +1,55 @@
+# SPDX-FileCopyrightText: 2025 OmniNode.ai Inc.
+# SPDX-License-Identifier: MIT
+"""Eval result model for autonomous off-peak evaluations.
+
+Related:
+    - OMN-6795: Define eval task models and enums
+    - OMN-6796: Build eval runner service
+
+.. versionadded:: 0.29.0
+"""
+
+from __future__ import annotations
+
+from uuid import UUID
+
+from pydantic import BaseModel, ConfigDict, Field
+
+from omnibase_infra.models.eval.model_eval_finding import ModelEvalFinding
+
+
+class ModelEvalResult(BaseModel):
+    """Result of an evaluation task execution.
+
+    Attributes:
+        task_id: The ID of the completed task.
+        findings: List of findings from the evaluation.
+        high_count: Number of high-severity findings.
+        medium_count: Number of medium-severity findings.
+        low_count: Number of low-severity findings.
+        tokens_used: Total tokens consumed.
+        actual_cost_usd: Actual cost in USD.
+        duration_ms: Execution duration in milliseconds.
+        llm_model_label: Human-readable label of model used.
+        llm_provider_label: Human-readable label of the provider that served the request.
+    """
+
+    model_config = ConfigDict(frozen=True, extra="forbid")
+
+    task_id: UUID = Field(..., description="Task identifier.")
+    findings: tuple[ModelEvalFinding, ...] = Field(
+        default_factory=tuple, description="Findings from the evaluation."
+    )
+    high_count: int = Field(default=0, ge=0, description="High-severity count.")
+    medium_count: int = Field(default=0, ge=0, description="Medium-severity count.")
+    low_count: int = Field(default=0, ge=0, description="Low-severity count.")
+    tokens_used: int = Field(default=0, ge=0, description="Total tokens consumed.")
+    actual_cost_usd: float = Field(default=0.0, ge=0.0, description="Actual cost.")
+    duration_ms: int = Field(default=0, ge=0, description="Execution duration in ms.")
+    llm_model_label: str = Field(default="", description="Model used.")
+    llm_provider_label: str = Field(
+        default="", description="Provider that served request."
+    )
+
+
+__all__: list[str] = ["ModelEvalResult"]

src/omnibase_infra/models/eval/model_eval_task.py

@@ -0,0 +1,63 @@
+# SPDX-FileCopyrightText: 2025 OmniNode.ai Inc.
+# SPDX-License-Identifier: MIT
+"""Eval task model for autonomous off-peak evaluations.
+
+Defines the task request model used by ``ServiceEvalRunner``
+to dispatch evaluation tasks to cheap LLM providers.
+
+Related:
+    - OMN-6795: Define eval task models and enums
+    - OMN-6796: Build eval runner service
+
+.. versionadded:: 0.29.0
+"""
+
+from __future__ import annotations
+
+from uuid import UUID, uuid4
+
+from pydantic import BaseModel, ConfigDict, Field
+
+from omnibase_infra.enums.enum_eval_task_status import EnumEvalTaskStatus
+from omnibase_infra.enums.enum_eval_task_type import EnumEvalTaskType
+
+
+class ModelEvalTask(BaseModel):
+    """An evaluation task to be dispatched to a cheap LLM provider.
+
+    Attributes:
+        task_id: Unique identifier for this task.
+        task_type: The type of evaluation to perform.
+        target_repo: Repository name to evaluate (e.g., "omnibase_infra").
+        target_path: Specific file or directory path to evaluate.
+        llm_model_label: Human-readable LLM model label for evaluation.
+        max_tokens: Maximum token budget for this individual task.
+        estimated_cost_usd: Estimated cost in USD for this task.
+        status: Current task status.
+    """
+
+    model_config = ConfigDict(frozen=True, extra="forbid")
+
+    task_id: UUID = Field(default_factory=uuid4, description="Unique task identifier.")
+    task_type: EnumEvalTaskType = Field(
+        ..., description="Type of evaluation to perform."
+    )
+    target_repo: str = Field(..., min_length=1, description="Repository to evaluate.")
+    target_path: str = Field(
+        default="", description="File or directory path within the repo."
+    )
+    llm_model_label: str = Field(
+        default="", description="LLM model to use (empty = auto-select via bifrost)."
+    )
+    max_tokens: int = Field(
+        default=4096, ge=256, le=131072, description="Max tokens for this task."
+    )
+    estimated_cost_usd: float = Field(
+        default=0.0, ge=0.0, description="Estimated cost in USD."
+    )
+    status: EnumEvalTaskStatus = Field(
+        default=EnumEvalTaskStatus.PENDING, description="Task status."
+    )
+
+
+__all__: list[str] = ["ModelEvalTask"]