devflow-engine 1.0.0__py3-none-any.whl

devflow_engine/prompts/integration/red_review/past_prompts/20260417T212300/red_review/prompt.md

@@ -0,0 +1,15 @@
+# Red Review
+
+RedReview is a real GenAI node: critique red integration proofs for alignment, adequacy, and workflow-side-effect coverage.
+
+- This is a real GenAI DAG. Return JSON only.
+- Node 1 and Node 2 must be genuinely model-backed; do not simulate agentic behavior with deterministic shortcuts.
+- Workflows are anchored to side effects, not users.
+- Attach implicated users to workflows using the canonical actor model: kind x scope x authority.
+- Reject denied canonical actor combinations by listing them under denied_candidates with explicit reasons.
+- Treat interaction points as first-class. Reference implemented seams when they exist; mark needed interaction points when required but missing.
+- Preserve alignment to implemented stories, implementation evidence, and source docs.
+- Do not overclaim unsupported implementation. Keep rationale explicit.
+- Review every red package against its workflow and side effect.
+- Find overreach, under-assertion, poor interaction-point anchoring, and missing actor-route coverage.
+- Return reviewed_packages as the red packages after critique, plus findings and repair_directions per workflow.

devflow_engine/prompts/integration/red_review/red_review/prompt.md

@@ -0,0 +1,9 @@
+# Red Review
+
+Review red integration proof packages for alignment, adequacy, and coverage.
+
+- Review every red package against its workflow and side effect.
+- Find overreach, under-assertion, weak interaction-point anchoring, and missing actor-route coverage.
+- Return reviewed_packages plus findings and repair_directions per workflow.
+- Keep critiques evidence-based; do not invent unsupported failures.
+- Return JSON only.

devflow_engine/prompts/integration/resolve/example.md

@@ -0,0 +1,111 @@
+# Resolve node assembled payloads
+
+## Provenance
+
+This file is **reconstructed from code**, not copied from a captured integration agent run.
+
+Concrete sources used:
+
+- stage prompt bullets:
+  - `prompts/integration/resolve/resolve_side_effects/prompt.md`
+  - `prompts/integration/resolve/resolve_implicated_users/prompt.md`
+- runtime context assembly:
+  - `src/devflow_engine/integration/dag.py` (`run_integration_agent_step(... context_payload=...)` for resolve)
+- prompt envelope assembly:
+  - `src/devflow_engine/agentic_runtime.py` → `_build_prompt(...)`
+- output schema source:
+  - `src/devflow_engine/integration/dag.py` → `SideEffectsArtifact`, `ImplicatedUsersArtifact`
+
+## Stable doctrine vs actual payload
+
+- Stable node doctrine file: `prompts/integration/resolve/node_config/prompt.md`
+- Actual `instructions` in the model payload come from the stage prompt files above.
+- Actual `context` comes from `integration/dag.py`.
+- Current integration path injects no skills and no tools here.
+
+## Assembled payload: `integration_resolve_side_effects`
+
+```json
+{
+  "task": "integration_resolve_side_effects",
+  "instructions": [
+    "Use only supported evidence from the implemented idea, implemented stories, code evidence, and source docs.",
+    "Return every side effect the implementation actually accomplishes.",
+    "For each side effect, include concise rationale and supported story_ids.",
+    "Include concrete interaction_points when implemented; otherwise use needed_interaction_points.",
+    "Populate alignment_context with brief counts or notes that help downstream validation.",
+    "Return JSON only."
+  ],
+  "context": {
+    "idea_id": "idea_checkout_recovery_v1",
+    "implemented_idea": {
+      "idea_id": "idea_checkout_recovery_v1",
+      "title": "Recover abandoned checkout carts",
+      "summary": "Detect abandoned carts and send recovery reminders when consent is present."
+    },
+    "implemented_stories": [
+      {
+        "story_id": "STORY-cart-abandonment-detection",
+        "title": "Detect abandoned carts",
+        "side_effect_ids": ["se_cart_marked_abandoned"]
+      },
+      {
+        "story_id": "STORY-recovery-email-send",
+        "title": "Send recovery reminder email",
+        "side_effect_ids": ["se_recovery_email_sent"]
+      }
+    ],
+    "code_evidence": [
+      { "path": "src/cart/recovery.py", "side_effect_ids": [], "interaction_points": [] },
+      { "path": "src/jobs/recovery_email_job.py", "side_effect_ids": [], "interaction_points": [] }
+    ],
+    "source_docs": [
+      { "path": "ai_docs/context/v2/project_docs/user_workflows.md" },
+      { "path": "ai_docs/context/v2/project_docs/domain_entities.md" }
+    ],
+    "canonical_actor_allowed_combinations": [
+      ["human", "customer_account", "self"],
+      ["human", "organization_workspace", "admin"],
+      ["system", "organization_workspace", "service"],
+      ["system", "platform", "service"]
+    ]
+  },
+  "output_schema": "SideEffectsArtifact.model_json_schema() from src/devflow_engine/integration/dag.py",
+  "return_format": "json_only"
+}
+```
+
+## Assembled payload: `integration_resolve_implicated_users`
+
+Same runtime context fields, but different stage prompt bullets and different output schema:
+
+```json
+{
+  "task": "integration_resolve_implicated_users",
+  "instructions": [
+    "Use only supported evidence from the implemented idea, implemented stories, code evidence, and source docs.",
+    "Resolve implicated users into the canonical actor model only: kind, scope, authority, rationale.",
+    "Only emit canonical actor combinations allowed by the provided context.",
+    "Put rejected combinations in denied_candidates with explicit reasons.",
+    "Do not invent story-local role labels or unsupported actors.",
+    "Return JSON only."
+  ],
+  "context": "same field set and sources as integration_resolve_side_effects",
+  "output_schema": "ImplicatedUsersArtifact.model_json_schema() from src/devflow_engine/integration/dag.py",
+  "return_format": "json_only"
+}
+```
+
+## Transport/runtime config that is separate from the prompt body
+
+At `run_agent_step(...)` call time:
+
+- `delivery_model="streaming"`
+- `interaction_model="agentic"`
+- `response_contract="json_only"`
+
+If Anthropic API transport is selected later by `invoke_llm(...)`, the HTTP payload uses:
+
+- `system_prompt=""`
+- one user message containing the JSON prompt above
+- `tools=[]`

devflow_engine/prompts/integration/resolve/node_config/prompt.md

@@ -0,0 +1,64 @@
+# Resolve Node Contract
+
+Identify implemented side effects and implicated canonical actors from the provided evidence.
+
+## What this file is
+
+This file is the stable node doctrine for the Resolve node.
+
+It is loaded by:
+
+- `src/devflow_engine/integration/prompts.py` → `load_integration_node_instruction("resolve")`
+- `src/devflow_engine/integration/dag.py` → `ResolveSideEffectsAndImplicatedUsersNode.get_agent_config()`
+
+## What this file is not
+
+It is **not** the literal prompt body sent to the model by the current integration runtime.
+
+The current model-facing prompt is assembled from:
+
+- stage prompt bullets in:
+  - `prompts/integration/resolve/resolve_side_effects/prompt.md`
+  - `prompts/integration/resolve/resolve_implicated_users/prompt.md`
+- runtime `context_payload` built in `src/devflow_engine/integration/dag.py`
+- `output_schema` injected by `src/devflow_engine/agentic_runtime.py`
+- `return_format: "json_only"` injected by `src/devflow_engine/agentic_runtime.py`
+
+## Real injected runtime components for this node
+
+### Resolve side effects stage
+
+`context_payload` fields:
+
+- `idea_id` ← `event.idea_id`
+- `implemented_idea` ← `event.implemented_idea`
+- `implemented_stories` ← `event.implemented_stories`
+- `code_evidence` ← `event.code_evidence`
+- `source_docs` ← `event.source_docs`
+- `canonical_actor_allowed_combinations` ← sorted `CANONICAL_ACTOR_ALLOWED_COMBINATIONS`
+
+`instructions` source:
+
+- bullets parsed from `prompts/integration/resolve/resolve_side_effects/prompt.md`
+
+`output_schema` source:
+
+- `SideEffectsArtifact.model_json_schema()`
+
+### Resolve implicated users stage
+
+Same `context_payload` sources as above.
+
+`instructions` source:
+
+- bullets parsed from `prompts/integration/resolve/resolve_implicated_users/prompt.md`
+
+`output_schema` source:
+
+- `ImplicatedUsersArtifact.model_json_schema()`
+
+## Tools / skills / config
+
+- No integration-specific skills are injected here.
+- No tool definitions are assembled here.
+- Runtime model/provider selection is handled separately by `src/devflow_engine/llm/invoke.py`.

devflow_engine/prompts/integration/resolve/past_prompts/20260417T212300/node_config/prompt.md

@@ -0,0 +1,64 @@
+# Resolve Node Contract
+
+Resolve implemented idea side effects and implicated canonical actors from implementation evidence.
+
+## What this file is
+
+This file is the stable node doctrine for the Resolve node.
+
+It is loaded by:
+
+- `src/devflow_engine/integration/prompts.py` → `load_integration_node_instruction("resolve")`
+- `src/devflow_engine/integration/dag.py` → `ResolveSideEffectsAndImplicatedUsersNode.get_agent_config()`
+
+## What this file is not
+
+It is **not** the literal prompt body sent to the model by the current integration runtime.
+
+The current model-facing prompt is assembled from:
+
+- stage prompt bullets in:
+  - `prompts/integration/resolve/resolve_side_effects/prompt.md`
+  - `prompts/integration/resolve/resolve_implicated_users/prompt.md`
+- runtime `context_payload` built in `src/devflow_engine/integration/dag.py`
+- `output_schema` injected by `src/devflow_engine/agentic_runtime.py`
+- `return_format: "json_only"` injected by `src/devflow_engine/agentic_runtime.py`
+
+## Real injected runtime components for this node
+
+### Resolve side effects stage
+
+`context_payload` fields:
+
+- `idea_id` ← `event.idea_id`
+- `implemented_idea` ← `event.implemented_idea`
+- `implemented_stories` ← `event.implemented_stories`
+- `code_evidence` ← `event.code_evidence`
+- `source_docs` ← `event.source_docs`
+- `canonical_actor_allowed_combinations` ← sorted `CANONICAL_ACTOR_ALLOWED_COMBINATIONS`
+
+`instructions` source:
+
+- bullets parsed from `prompts/integration/resolve/resolve_side_effects/prompt.md`
+
+`output_schema` source:
+
+- `SideEffectsArtifact.model_json_schema()`
+
+### Resolve implicated users stage
+
+Same `context_payload` sources as above.
+
+`instructions` source:
+
+- bullets parsed from `prompts/integration/resolve/resolve_implicated_users/prompt.md`
+
+`output_schema` source:
+
+- `ImplicatedUsersArtifact.model_json_schema()`
+
+## Tools / skills / config
+
+- No integration-specific skills are injected here.
+- No tool definitions are assembled here.
+- Runtime model/provider selection is handled separately by `src/devflow_engine/llm/invoke.py`.

devflow_engine/prompts/integration/resolve/past_prompts/20260417T212300/resolve_implicated_users/prompt.md

@@ -0,0 +1,15 @@
+# Resolve Implicated Users
+
+Resolve implemented idea side effects and implicated canonical actors from implementation evidence.
+
+- This is a real GenAI DAG. Return JSON only.
+- Node 1 and Node 2 must be genuinely model-backed; do not simulate agentic behavior with deterministic shortcuts.
+- Workflows are anchored to side effects, not users.
+- Attach implicated users to workflows using the canonical actor model: kind x scope x authority.
+- Reject denied canonical actor combinations by listing them under denied_candidates with explicit reasons.
+- Treat interaction points as first-class. Reference implemented seams when they exist; mark needed interaction points when required but missing.
+- Preserve alignment to implemented stories, implementation evidence, and source docs.
+- Do not overclaim unsupported implementation. Keep rationale explicit.
+- Resolve implicated users into the canonical actor model only: kind, scope, authority, rationale.
+- Reject non-listed canonical actor combinations into denied_candidates with explicit reasons.
+- Do not invent story-local role labels as canonical actor classes.

devflow_engine/prompts/integration/resolve/past_prompts/20260417T212300/resolve_side_effects/prompt.md

@@ -0,0 +1,15 @@
+# Resolve Side Effects
+
+Resolve implemented idea side effects and implicated canonical actors from implementation evidence.
+
+- This is a real GenAI DAG. Return JSON only.
+- Node 1 and Node 2 must be genuinely model-backed; do not simulate agentic behavior with deterministic shortcuts.
+- Workflows are anchored to side effects, not users.
+- Attach implicated users to workflows using the canonical actor model: kind x scope x authority.
+- Reject denied canonical actor combinations by listing them under denied_candidates with explicit reasons.
+- Treat interaction points as first-class. Reference implemented seams when they exist; mark needed interaction points when required but missing.
+- Preserve alignment to implemented stories, implementation evidence, and source docs.
+- Do not overclaim unsupported implementation. Keep rationale explicit.
+- Identify the implemented idea's real accomplished outcomes as side effects.
+- Return every side effect with rationale, story_ids when supported, and interaction_points or needed_interaction_points where evidence supports them.
+- Populate alignment_context with counts or notes that help downstream validation.

devflow_engine/prompts/integration/resolve/resolve_implicated_users/prompt.md

@@ -0,0 +1,10 @@
+# Resolve Implicated Users
+
+Identify implicated canonical actors from the provided evidence.
+
+- Use only supported evidence from the implemented idea, implemented stories, code evidence, and source docs.
+- Resolve implicated users into the canonical actor model only: kind, scope, authority, rationale.
+- Only emit canonical actor combinations allowed by the provided context.
+- Put rejected combinations in denied_candidates with explicit reasons.
+- Do not invent story-local role labels or unsupported actors.
+- Return JSON only.

devflow_engine/prompts/integration/resolve/resolve_side_effects/prompt.md

@@ -0,0 +1,10 @@
+# Resolve Side Effects
+
+Identify implemented side effects from the provided evidence.
+
+- Use only supported evidence from the implemented idea, implemented stories, code evidence, and source docs.
+- Return every side effect the implementation actually accomplishes.
+- For each side effect, include concise rationale and supported story_ids.
+- Include concrete interaction_points when implemented; otherwise use needed_interaction_points.
+- Populate alignment_context with brief counts or notes that help downstream validation.
+- Return JSON only.

devflow_engine/prompts/integration/validate/build_idea_acceptance_coverage/prompt.md

@@ -0,0 +1,12 @@
+# VEG Idea Acceptance Coverage Builder
+
+Build an evidence-backed idea-acceptance coverage artifact in VEG and return workflows ready for deterministic gating.
+
+- Return exactly one enriched_workflow per input workflow and preserve every workflow_id + side-effect anchor.
+- Build idea_acceptance_coverage with exactly one criterion entry per implemented IDEA acceptance criterion.
+- For each criterion, map the supporting story_ids, workflow_ids, side_effect_ids, evidence_refs, and a concrete proof_summary.
+- Mark verdict as covered only when the implemented stories collectively prove the criterion through real workflow seams and cited evidence.
+- If proof is weak, missing, or seamless, mark the criterion partial or uncovered and explain the exact gaps / missing_seams. Do not soft-pass.
+- protected_sections and do_not_touch should preserve already-proven criteria/workflows so later repair stays scoped.
+- Make unsupported or inconsistent workflow claims explicit in findings instead of inventing facts.
+- Return JSON only.

devflow_engine/prompts/integration/validate/code_repair/prompt.md

@@ -0,0 +1,13 @@
+# Validate Code Repair
+
+Repair product code for the current validation failure.
+
+- Patch the smallest real code change that fixes the current failing workflow surface.
+- Choose one primary failing seam from validation_errors, failing_workflows, repair_sources, and code_evidence_files; keep edits there unless the same failure forces one adjacent file.
+- Treat protected_sections as immutable context and obey every item in do_not_touch.
+- Prefer files already named in code_evidence_files, workflow backing, or the failing interaction seam.
+- Update updated_workflows only when the repaired code changes workflow truth.
+- Do not repair red-package artifacts here. Do not edit engine, runtime, orchestration, or prompt-loading layers unless the failure explicitly originates there.
+- If the issue is workflow-only, artifact-only, or runtime-only, leave code unchanged and explain it in unresolvable_failures.
+- Return only changed files with full contents, updated_workflows, and unresolvable_failures.
+- Return JSON only.

devflow_engine/prompts/integration/validate/example.md

@@ -0,0 +1,143 @@
+# Validate node assembled payloads
+
+## Provenance
+
+Reconstructed from:
+
+- `prompts/integration/validate/validate_enrich_gate/prompt.md`
+- `prompts/integration/validate/code_repair/prompt.md`
+- `src/devflow_engine/integration/dag.py` validate gate + code repair calls
+- `src/devflow_engine/agentic_runtime.py`
+- `ValidateEnrichArtifact` and `CodeRepairArtifact` in `src/devflow_engine/integration/dag.py`
+
+Not copied from a persisted past-run integration prompt.
+
+## Stable doctrine vs actual payload
+
+- Stable node doctrine file: `prompts/integration/validate/node_config/prompt.md`
+- Actual `instructions` come from stage prompt markdown plus the runtime repair-iteration line on repair passes
+- Actual `context` comes from validate-time artifacts and repair-time derived data in `integration/dag.py`
+- No skills or tools are injected here
+
+## Assembled payload: `integration_validate_enrich_gate`
+
+```json
+{
+  "task": "integration_validate_enrich_gate",
+  "instructions": [
+    "Review every workflow and return exactly one enriched_workflow per input workflow.",
+    "Preserve workflow identity and side-effect anchoring.",
+    "Improve rationale, backing, sequencing, branches, resulting artifacts, and interaction-point descriptions only where the evidence supports it.",
+    "Make unsupported, inconsistent, or under-supported areas explicit in findings.",
+    "Do not invent seams, actors, or implementation facts.",
+    "Return JSON only."
+  ],
+  "context": {
+    "idea_id": "idea_checkout_recovery_v1",
+    "side_effects_artifact": {
+      "idea_id": "idea_checkout_recovery_v1",
+      "side_effects": [
+        {
+          "id": "se_recovery_email_sent",
+          "interaction_points": [
+            { "kind": "endpoint", "name": "POST /api/carts/{cart_id}/recover", "implemented": true }
+          ]
+        }
+      ]
+    },
+    "implicated_users_artifact": {
+      "idea_id": "idea_checkout_recovery_v1",
+      "implicated_users": [
+        { "kind": "human", "scope": "customer_account", "authority": "self", "rationale": "customer receives reminder" }
+      ],
+      "denied_candidates": []
+    },
+    "workflows": [
+      {
+        "workflow_id": "wf_recovery_email_sent",
+        "idea_id": "idea_checkout_recovery_v1",
+        "side_effect": { "id": "se_recovery_email_sent", "name": "Recovery email sent" },
+        "implicated_users": [
+          { "kind": "human", "scope": "customer_account", "authority": "self" }
+        ],
+        "story_backing": [
+          { "story_id": "STORY-recovery-email-send", "title": "Send recovery reminder email" }
+        ],
+        "code_backing": [
+          { "path": "src/jobs/recovery_email_job.py" }
+        ],
+        "source_doc_backing": [
+          { "path": "ai_docs/context/v2/project_docs/user_workflows.md" }
+        ],
+        "interaction_points": [
+          { "kind": "endpoint", "name": "POST /api/carts/{cart_id}/recover", "implemented": true }
+        ],
+        "needed_interaction_points": [],
+        "process_sequence": ["cart abandoned", "job scheduled", "email sent"],
+        "branches": [],
+        "resulting_artifacts": ["queued recovery email job"],
+        "mermaid": null
+      }
+    ],
+    "implemented_stories": [
+      { "story_id": "STORY-recovery-email-send", "title": "Send recovery reminder email", "side_effect_ids": ["se_recovery_email_sent"], "required_planes": [] }
+    ],
+    "code_evidence": [
+      { "path": "src/jobs/recovery_email_job.py", "side_effect_ids": [], "interaction_points": [] }
+    ],
+    "source_docs": [
+      { "path": "ai_docs/context/v2/project_docs/user_workflows.md" }
+    ]
+  },
+  "output_schema": "ValidateEnrichArtifact.model_json_schema()",
+  "return_format": "json_only"
+}
+```
+
+## Assembled payload: `integration_code_repair_iter1` (validate repair mode)
+
+```json
+{
+  "task": "integration_code_repair_iter1",
+  "instructions": [
+    "Patch the smallest real code change that fixes the current failing workflow surface.",
+    "Choose one primary failing seam from validation_errors, failing_workflows, and code_evidence_files; keep edits there unless the same failure forces one adjacent file.",
+    "Prefer files already named in code_evidence_files, workflow backing, or the failing interaction seam.",
+    "Update updated_workflows only when the repaired code changes workflow truth.",
+    "Do not repair red-package artifacts here. Do not edit engine, runtime, orchestration, or prompt-loading layers unless the failure explicitly originates there.",
+    "If the issue is workflow-only, artifact-only, or runtime-only, leave code unchanged and explain it in unresolvable_failures.",
+    "Return only changed files with full contents, updated_workflows, and unresolvable_failures.",
+    "This is repair iteration 1. Stay on the current failing seam. Do not revisit clean surfaces.",
+    "Return JSON only."
+  ],
+  "context": {
+    "failing_workflows": [
+      {
+        "workflow_id": "wf_recovery_email_sent",
+        "side_effect": { "id": "se_recovery_email_sent", "name": "Recovery email sent" },
+        "needed_interaction_points": [
+          { "kind": "endpoint", "name": "POST /api/carts/{cart_id}/recover" }
+        ]
+      }
+    ],
+    "validation_errors": [
+      "wf_recovery_email_sent: MISSING_INTERACTION_POINTS endpoint route not declared"
+    ],
+    "code_evidence_files": [
+      {
+        "path": "src/cart/recovery.py",
+        "content": "def recover_cart(cart_id: str):
+    queue_recovery_email(cart_id)
+"
+      }
+    ],
+    "side_effects_artifact": { "idea_id": "idea_checkout_recovery_v1", "side_effects": [{ "id": "se_recovery_email_sent" }] },
+    "implicated_users_artifact": { "idea_id": "idea_checkout_recovery_v1", "implicated_users": [{ "kind": "human", "scope": "customer_account", "authority": "self", "rationale": "customer receives reminder" }], "denied_candidates": [] },
+    "implemented_stories": [
+      { "story_id": "STORY-recovery-email-send", "title": "Send recovery reminder email", "side_effect_ids": ["se_recovery_email_sent"], "required_planes": [] }
+    ]
+  },
+  "output_schema": "CodeRepairArtifact.model_json_schema()",
+  "return_format": "json_only"
+}
+```

devflow_engine/prompts/integration/validate/node_config/prompt.md

@@ -0,0 +1,87 @@
+# Validate / VEG Node Contract
+
+Agentic VEG builder + deterministic gate for idea-acceptance validation.
+
+## What this file is
+
+Stable node doctrine loaded by:
+
+- `load_integration_node_instruction("validate")`
+- `BuildIdeaAcceptanceCoverageNode.get_agent_config()`
+
+## What actually reaches the model
+
+The current integration runtime uses two validate-area model-facing prompt shapes:
+
+1. VEG idea-acceptance builder (`integration_build_idea_acceptance_coverage`)
+2. validate repair pass (`integration_validate_repair`)
+
+The checker/gate itself is deterministic Python (`ValidateEnrichGateNode`) and is not model-backed.
+
+## Real injected runtime components
+
+### VEG idea-acceptance builder
+
+`context_payload` fields:
+
+- `idea_id`
+- `implemented_idea`
+- `implemented_stories`
+- `side_effects_artifact`
+- `implicated_users_artifact`
+- `workflows`
+- `code_evidence`
+- `source_docs`
+
+`instructions` source:
+
+- bullets parsed from `prompts/integration/validate/build_idea_acceptance_coverage/prompt.md`
+
+`output_schema` source:
+
+- `VegIdeaAcceptanceBuilderArtifact.model_json_schema()`
+
+### Validate repair pass
+
+`context_payload` fields:
+
+- `enriched_workflows`
+- `idea_acceptance_coverage`
+- `failing_workflows`
+- `failing_idea_acceptance_criteria`
+- `deterministic_errors`
+- `repair_sources`
+- `protected_sections`
+- `do_not_touch`
+
+`instructions` source:
+
+- bullets parsed from `prompts/integration/validate/validate_repair/prompt.md`
+
+`output_schema` source:
+
+- `VegIdeaAcceptanceBuilderArtifact.model_json_schema()`
+
+### Validate code repair pass
+
+`context_payload` fields:
+
+- `failing_workflows`
+- `validation_errors`
+- `repair_sources`
+- `protected_sections`
+- `do_not_touch`
+- `code_evidence_files`
+- `side_effects_artifact`
+- `implicated_users_artifact`
+- `implemented_stories`
+- `idea_acceptance_coverage`
+- `failing_idea_acceptance_criteria`
+
+`instructions` source:
+
+- bullets parsed from `prompts/integration/validate/code_repair/prompt.md`
+
+`output_schema` source:
+
+- `CodeRepairArtifact.model_json_schema()`

devflow_engine/prompts/integration/validate/past_prompts/20260417T212300/code_repair/prompt.md

@@ -0,0 +1,19 @@
+# Validate Code Repair
+
+You are the Integration Code Repair agent. Validation failed. Repair the code so it passes.
+
+- This is a real GenAI DAG. Return JSON only.
+- Node 1 and Node 2 must be genuinely model-backed; do not simulate agentic behavior with deterministic shortcuts.
+- Workflows are anchored to side effects, not users.
+- Attach implicated users to workflows using the canonical actor model: kind x scope x authority.
+- Reject denied canonical actor combinations by listing them under denied_candidates with explicit reasons.
+- Treat interaction points as first-class. Reference implemented seams when they exist; mark needed interaction points when required but missing.
+- Preserve alignment to implemented stories, implementation evidence, and source docs.
+- Do not overclaim unsupported implementation. Keep rationale explicit.
+- You are the Integration Code Repair agent. Validation failed. Repair the code so it passes.
+- For MISSING_INTERACTION_POINTS: inspect the code evidence for the side effect, find or create the actual seam interfaces, write the code. Declare the interaction_points in the updated workflow.
+- For STORY_ALIGNMENT_MISMATCH: inspect the code evidence against the story contract. Patch the code so it satisfies the story intent. Update the workflow to correctly describe the implementation.
+- For MISSING_CODE_BACKING: implement the missing code seam for the side effect. Write real implementation code, not stubs.
+- Write only the files that need to change. Return full file content for each patched file.
+- Return updated_workflows with corrected interaction_points and story_backing for every workflow you repaired.
+- In unresolvable_failures list any failure you cannot fix with code changes (e.g. fundamentally undefined story contract).