# Sovereign Executive Interface — Full Agent Brief

> Provable, local-first intent router for autonomous enterprise operations.

SEI translates high-level executive intent into validated, evidence-cited, multi-role agentic execution. Every worker output is verified by an independent Guardian against published standards and recorded in a tamper-evident, hash-chained provability ledger. Default execution is local-first, so client data need never leave client hardware.

## How it works
Intent -> validated execution graph -> Sovereign Role dispatch -> independent Guardian validation -> verified, evidence-cited outcome. Every model call is recorded in a tamper-evident, hash-chained ledger.

## Sovereign Roles
### Sovereign_Compliance (`sovereign_compliance`)
Jurisdictional mapping, regulatory auditing, contract and sanctions review, risk mitigation.
_Routing signals: legal, compliance, regulation, regulatory, audit, sanction, jurisdiction, contract_

### Sovereign_Crisis (`sovereign_crisis`)
Crisis command: incident triage, containment sequencing, stakeholder communication, business continuity, post-incident hardening.
_Routing signals: crisis, incident, breach, outage, recall, emergency, ransomware, continuity_

### Sovereign_Deals (`sovereign_deals`)
Deal architecture: M&A and investment analysis, valuation framing, term-sheet structuring, due-diligence orchestration, negotiation strategy.
_Routing signals: acquisition, merger, m&a, deal, term sheet, valuation, due diligence, negotiation_

### Sovereign_Supply_Chain (`sovereign_supply_chain`)
Logistics network analysis, route and carrier alternatives, inventory positioning, disruption response.
_Routing signals: logistics, shipment, shipping, route, freight, port, supplier, supply_

### Sovereign_Tax (`sovereign_tax`)
Global tax architecture: entity structuring, transfer pricing, treaty analysis, withholding optimization, nexus and domicile strategy.
_Routing signals: tax, vat, gst, transfer pricing, treaty, withholding, domicile, nexus_

### Sovereign_Treasury (`sovereign_treasury`)
Liquidity monitoring, cash flow and financial exposure analysis, yield optimization, demand-leakage detection and recovery.
_Routing signals: cash, liquidity, payment, invoice, leakage, treasury, cost, budget_

### Sovereign_Intelligence (`sovereign_intelligence`)
Market and competitive intelligence: competitor mapping, pricing landscapes, positioning analysis, trend and threat detection.
_Routing signals: competitor, competitive, market, intelligence, pricing, landscape, positioning, trend_

### Sovereign_Talent (`sovereign_talent`)
Workforce planning, talent acquisition strategy, human-AI hybrid role mapping, compensation benchmarking.
_Routing signals: talent, hiring, hire, workforce, team, headcount, compensation, recruit_

### Sovereign_Analyst (`sovereign_analyst`)
General research, synthesis, and analysis for intents no specialist role covers. The routing fallback.
_Routing signals: general_

## Pricing
- Model: outcome_based_performance_fee
- No recovery, no fee. 20% of independently verified recovered or created value.
- Basis: The fee applies only to value the Guardian validation loop and the hash-chained provability ledger confirm as recovered or created. Disputes are settled by cited evidence, not assertion.
- Example: $100,000 recovered -> $20,000 fee, client keeps $80,000.

## Trust guarantees
- Every run is recorded in a SHA-256 hash-chained ledger; any edit, insertion, or deletion is detectable with `sei verify <run>`.
- No worker output reaches the client unless an independent Guardian agent verifies it against published standards (the Puck Standard).
- Local-first and air-gappable. Default execution uses on-box models; prompts and client data need never leave the client's hardware.
- A human approval gate governs irreversible or outward-facing actions; the system asks before it acts beyond analysis.

## Security
- Evidence refs are mechanically verified before any model vote.
- A Guardian that cannot complete its audit rejects rather than approves.
- Model-call ledger is tamper-evident (hash-chained).

## Agent interface
- MCP tools: describe, get_offer, plan_intent, verify_run, list_runs
- Auth (advertised): none
- Rate limit: 60/min