worflow enhancements

Author: Zvi Fried

src/mcp_as_a_judge/models.py

@@ -259,6 +259,28 @@ class JudgeCodingPlanUserVars(BaseModel):
         description="Previous conversation history as JSON array with timestamps",
     )
 
+    # Domain focus and reuse/dependency planning (explicit inputs so plans don't get rejected)
+    problem_domain: str = Field(
+        default="",
+        description="Concise statement of the problem domain being solved",
+    )
+    problem_non_goals: list[str] = Field(
+        default_factory=list,
+        description="Explicit non-goals/out-of-scope items to prevent scope creep",
+    )
+    # Library selection map and internal reuse map are free-form lists of objects
+    # Using list[dict] to keep input flexible for different ecosystems
+    library_plan: list[dict] = Field(
+        default_factory=list,
+        description=(
+            "List of {purpose, selection, source: internal|external|custom, justification}"
+        ),
+    )
+    internal_reuse_components: list[dict] = Field(
+        default_factory=list,
+        description="List of {path, purpose, notes} for reusable repo components",
+    )
+
     # Conditional research fields
     research_required: bool = Field(
         default=False, description="Whether external research is required for this task"

src/mcp_as_a_judge/prompts/tool_descriptions/judge_coding_plan.md

@@ -21,6 +21,10 @@ Validate a proposed plan and design against requirements, research needs, and ri
 - `research`: string — Findings and rationale (provide if available)
 - `research_urls`: list[string] — URLs for external research (if required)
 - `context`: string — Additional project context
+ - `problem_domain`: string — Concise problem domain statement (optional but recommended)
+ - `problem_non_goals`: list[string] — Non-goals/out-of-scope items (optional)
+ - `library_plan`: list[object] — Library Selection Map entries: {purpose, selection, source: internal|external|custom, justification}
+ - `internal_reuse_components`: list[object] — Internal Reuse Map entries: {path, purpose, notes}
 
 ## Returns
 - Response JSON schema (JudgeResponse):

src/mcp_as_a_judge/prompts/user/judge_coding_plan.md

@@ -19,6 +19,32 @@ Please evaluate the following coding plan:
 
 {{ design }}
 
+## Problem Domain Statement
+
+{{ problem_domain or "" }}
+
+{% if problem_non_goals %}
+### Non-Goals
+{% for item in problem_non_goals %}- {{ item }}
+{% endfor %}
+{% endif %}
+
+{% if library_plan %}
+## Library Selection Map (Purpose → Selection)
+
+```
+{{ library_plan }}
+```
+{% endif %}
+
+{% if internal_reuse_components %}
+## Internal Reuse Map (Repo Components)
+
+```
+{{ internal_reuse_components }}
+```
+{% endif %}
+
 ## Research
 
 {{ research|default("") }}

src/mcp_as_a_judge/server.py

@@ -1178,6 +1178,10 @@ async def _evaluate_coding_plan(
     conversation_history: list[dict],
     task_metadata: TaskMetadata,
     ctx: Context,
+    problem_domain: str | None = None,
+    problem_non_goals: list[str] | None = None,
+    library_plan: list[dict] | None = None,
+    internal_reuse_components: list[dict] | None = None,
 ) -> JudgeResponse:
     """Evaluate coding plan using AI judge.
 
@@ -1216,6 +1220,11 @@ async def _evaluate_coding_plan(
         else False,
         identified_risks=task_metadata.identified_risks or [],
         risk_mitigation_strategies=task_metadata.risk_mitigation_strategies or [],
+        # Domain focus and reuse maps (optional explicit inputs)
+        problem_domain=problem_domain or "",
+        problem_non_goals=problem_non_goals or [],
+        library_plan=library_plan or [],
+        internal_reuse_components=internal_reuse_components or [],
     )
     messages = create_separate_messages(
         "system/judge_coding_plan.md",
@@ -1252,6 +1261,11 @@ async def judge_coding_plan(
     context: str = "",
     # OPTIONAL override
     user_requirements: str | None = None,
+    # OPTIONAL explicit inputs to avoid rejection on missing deliverables
+    problem_domain: str | None = None,
+    problem_non_goals: list[str] | None = None,
+    library_plan: list[dict] | None = None,
+    internal_reuse_components: list[dict] | None = None,
 ) -> JudgeResponse:
     """Coding plan evaluation tool - description loaded from tool_description_provider."""
     # Log tool execution start
@@ -1265,6 +1279,10 @@ async def judge_coding_plan(
         "research": research,
         "context": context,
         "research_urls": research_urls,
+        "problem_domain": problem_domain,
+        "problem_non_goals": problem_non_goals,
+        "library_plan": library_plan,
+        "internal_reuse_components": internal_reuse_components,
     }
 
     try:
@@ -1522,6 +1540,10 @@ async def judge_coding_plan(
             history_json_array,
             task_metadata,  # Pass task metadata for conditional features
             ctx,
+            problem_domain=problem_domain,
+            problem_non_goals=problem_non_goals,
+            library_plan=library_plan,
+            internal_reuse_components=internal_reuse_components,
         )
 
         # Additional research validation if approved
@@ -1556,17 +1578,22 @@ async def judge_coding_plan(
         # If missing but the plan was approved, convert to required improvements
         missing_deliverables: list[str] = []
         try:
+            # Fill from explicit inputs if LLM omitted them in metadata
+            if problem_domain and not getattr(updated_task_metadata, "problem_domain", "").strip():
+                updated_task_metadata.problem_domain = problem_domain  # type: ignore[attr-defined]
+            if problem_non_goals and not getattr(updated_task_metadata, "problem_non_goals", None):
+                updated_task_metadata.problem_non_goals = problem_non_goals  # type: ignore[attr-defined]
+            if library_plan and (not getattr(updated_task_metadata, "library_plan", None) or len(getattr(updated_task_metadata, "library_plan", [])) == 0):
+                updated_task_metadata.library_plan = library_plan  # type: ignore[attr-defined]
+            if internal_reuse_components and (not getattr(updated_task_metadata, "internal_reuse_components", None) or len(getattr(updated_task_metadata, "internal_reuse_components", [])) == 0):
+                updated_task_metadata.internal_reuse_components = internal_reuse_components  # type: ignore[attr-defined]
+
+            # Now check for missing deliverables
             if not getattr(updated_task_metadata, "problem_domain", "").strip():
                 missing_deliverables.append("Add a clear Problem Domain Statement with explicit non-goals")
-            # library_plan is a list of LibraryPlanItem entries
-            if not getattr(updated_task_metadata, "library_plan", []) or len(
-                getattr(updated_task_metadata, "library_plan", [])
-            ) == 0:
-                missing_deliverables.append(
-                    "Provide a Library Selection Map (purpose → internal/external library with justification)"
-                )
+            if not getattr(updated_task_metadata, "library_plan", []) or len(getattr(updated_task_metadata, "library_plan", [])) == 0:
+                missing_deliverables.append("Provide a Library Selection Map (purpose → internal/external library with justification)")
         except Exception:
-            # Be resilient; do not crash on schema variations
             pass
 
         effective_approved = evaluation_result.approved and not missing_deliverables