mova-complaints-handler

> **Contract Skill** — A ready-to-use MOVA HITL workflow. Requires the `openclaw-mova` plugin. # MOVA EU Consumer Complaints Handler Turn any customer complaint into a structured, auditable handling decision — with AI triage, EU-compliant mandatory human review, and a signed audit receipt. ## Why use this skill - **AI classifies in seconds** — sentiment flags, product risk, regulator threat detection, repeat customer check - **Mandatory human gate** — compensation claims, regulator threats, high-risk products, and fraud signals always route to a human officer - **Full audit trail** — every decision is timestamped, signed, and stored in an immutable MOVA audit journal (EBA/ESMA-ready) - **Draft response hint** — AI suggests tone and content for the customer response ## Requirements **Plugin:** MOVA OpenClaw plugin must be installed in your OpenClaw workspace. ## Quick start Tell your agent: > "Handle complaint CMP-2026-1002 from customer CUST-992 — product: investments, text: 'I want compensation for unauthorized charges and will escalate to the regulator.', prior complaints: CMP-0501, CMP-0604" ## Demo    ## Test complaint [test_complaint_CMP-2026-1002.png](https://raw.githubusercontent.com/mova-compact/mova-bridge/main/test_complaint_CMP-2026-1002.png) ## What the agent does **Step 1 — Submit complaint** Call tool `mova_hitl_start_complaint` with: - `complaint_id` (e.g. CMP-2026-1002), `customer_id` - `complaint_text`, `channel` (web/email/phone/chat/branch) - `product_category` (e.g. payments, investments, mortgage) - `complaint_date` (ISO date) - `previous_complaints`: optional array of prior complaint IDs - `attachments`: optional array of filenames - `customer_segment`: optional (retail/sme/corporate) - `preferred_language`: optional ISO 639-1 code **Step 2 — AI analysis output** ``` Complaint: CMP-2026-1002 — investments Risk score: 0.85 / 1.0 (high) Sentiment flags: compensation_claim, regulator_threat Repeat customer: yes (2 prior complaints) Findings: • COMP_CLAIM (high) — Customer requests compensation for unauthorized charges • REGULATOR_THREAT (medium) — Customer threatens regulator escalation • PRODUCT_RISK_INVEST (high) — High-risk product category Draft response hint: Acknowledge receipt, confirm investigation, outline compensation process Recommended action: escalate ← RECOMMENDED ``` **Step 3 — Human decision gate** | Option | Description | |---|---| | `resolve` | Send standard response | | `escalate` | Forward to complaints officer ← recommended | | `reject` | Mark as incomplete / invalid | | `regulator_flag` | Flag for regulator reporting | Call tool `mova_hitl_decide` with: - `contract_id`: from the response above — this is `ctr-cmp-xxxxxxxx`, NOT the complaint ID - `option`: chosen decision - `reason`: officer reasoning **Step 4 — Audit receipt** Call tool `mova_hitl_audit` with `contract_id`. Call tool `mova_hitl_audit_compact` with `contract_id` for the full signed event chain. ## Audit log format (compact) | Seq | Event | Details | |---|---|---| | 1 | contract.start | Complaint submitted | | 2 | step.ai_task | Classification completed | | 3 | step.verification | Risk snapshot passed | | 4 | step.decision_point | Awaiting human | | 5 | decision.human | Officer selected option + reason | | 0 | meta | Contract finalized, status: completed | ## Data flows - Complaint text and metadata → MOVA Classification API (EU-hosted) - Human decision → MOVA Audit Journal (immutable, signed) - No data stored locally or sent to third parties ## Connect your real CRM and policy systems By default MOVA uses a sandbox mock. To route checks against your live CRM and policy engine, call `mova_list_connectors` with `keyword: "crm"`. Relevant connectors: | Connector ID | What it covers | |---|---| | `connector.crm.customer_lookup_v1` | Customer history and prior complaints from CRM | | `connector.policy.complaints_rules_v1` | Complaints handling rules by product/jurisdiction | | `connector.notification.email_v1` | Customer notification email | Call `mova_register_connector` with `connector_id`, `endpoint`, optional `auth_header` and `auth_value`. ## Rules - Agent never makes HTTP requests directly - Agent never invents or simulates results — if a tool call fails, show the exact error - Use MOVA plugin tools directly — do NOT use exec or shell - CONTRACT_ID is `ctr-cmp-xxxxxxxx` from the mova_hitl_start_complaint response — NOT the complaint ID