Missing Canonical Services Implementation Guide

Status: PROPOSED

Approved: NO

Execution Authority: NONE

Evidence Authority: POST_STOP_FORENSIC_RECONCILIATION.md

Original Status Claim: Services identified as missing during canonical restoration audit — UNAPPROVED PROPOSAL
Original Impact Claim: Customer Zero activation is blocked on these implementations — NOT VERIFIED AS AN EXECUTION PREREQUISITE

---

Missing Service 1: spine-writer

Purpose: Canonical write path for the Operational Spine
Current Workaround: integratewise-pipeline may be handling writes (legacy role)
Required Inputs:

• Entity mutations from workflows, actions, decisions
• Spine schema (entity types, relationships)
• Neon Fortress connection for canonical writes; D1/KV only for bounded
  operational state
• Continuity event publishing for audit

Required Outputs:

• Successful write confirmation (entity_id, version, timestamp)
• Continuity record created for recovery
• Timeline event published
• Cache invalidation for dependents

Implementation Steps:

1. Create services/spine-writer/src/index.ts
2. Implement entity write handler (POST /entities)
3. Connect the governed writer to Neon Fortress and emit bounded D1/KV
   invalidation/rehydration events
4. Publish continuity events
5. Deploy as integratewise-spine-writer

---

Missing Service 2: governance

Purpose: Policy evaluation, approval chains, governance decisions
Current Workaround: integratewise-govern (wrong name) may have partial logic
Required Inputs:

• Capability proposal or action request
• Policy configuration for tenant
• User/role context
• Historical decision records

Required Outputs:

• Governance decision (approved, rejected, requires_approval)
• Approval record created
• Policy audit trail

Implementation Steps:

1. Create services/governance/src/index.ts (note: correct spelling)
2. Load tenant policies from policy store
3. Evaluate against capability request
4. Route to approval chain if needed
5. Deploy as integratewise-governance

Note: Rename or deprecate integratewise-govern to avoid naming conflict.

---

Missing Service 3: schema-synthesis

Purpose: AI-driven schema inference and generation
Current Workaround: None (critical gap)
Required Inputs:

• Raw data samples (JSON, CSV, unstructured)
• Existing schema patterns (for similarity detection)
• User hints or type guidance

Required Outputs:

• Inferred schema (entity types, field mappings, relationships)
• Confidence scores
• Schema validation rules

Implementation Steps:

1. Create services/schema-synthesis/src/index.ts
2. Integrate HuggingFace inference for schema generation
3. Store inferred schemas in store service
4. Cache common patterns
5. Deploy as integratewise-schema-synthesis

---

Missing Service 4: huggingface-inference

Purpose: Inference endpoint for AI model calls
Current Workaround: Calls may be routed directly to HuggingFace (no abstraction)
Required Inputs:

• Model ID (from HuggingFace)
• Input prompt or data
• Parameters (temperature, max_tokens, etc.)

Required Outputs:

• Model output (text, embeddings, predictions)
• Inference latency
• Token usage for billing

Implementation Steps:

1. Create services/huggingface-inference/src/index.ts
2. Implement model inference handler
3. Support multiple model types (text generation, embeddings, classification)
4. Add caching layer for common queries
5. Deploy as integratewise-huggingface-inference

---

Timeline to Full Canonical Readiness

Service	Effort	Priority	Dependency
spine-writer	Medium	Critical	Needed for entity mutations
governance	Medium	High	Needed for approval chains
schema-synthesis	High	Medium	Needed for schema AI features
huggingface-inference	Low	High	Needed for all inference

Estimated total: 3–5 days for all four services

---

Customer Zero Activation Can Begin With:

• 12 deployed canonical services (currently live)
• Basic workflow execution (without governance approval chain)
• Data normalization and querying (without schema synthesis)
• Inference calls routed directly or through adapter

Full feature parity requires all 18 services.