Console Guide

The PhronEdge console at AI Constitutional Console is your governance command center. Sign in with Google, Microsoft, or GitHub.

Policy Builder

Build and sign constitutional policies for your AI agents.

Step 1: Define agents

Add each agent with:

• Purpose: what the agent does
• Use case: classification from the ACI engine (auto-scored)
• Model: which LLM the agent uses
• Tier: T0 (read-only) through T3 (orchestrator with delegation)
• Data classifications: what data levels the agent can access
• Tools: which functions the agent can call

Step 2: Add tools

For each agent, define the tools it may use:

• Tool name: must match the @pe.govern("tool_name") in your code
• Description: what the tool does
• Data classification: what data level the tool accesses
• Requires approval: whether a human must approve calls to this tool

Step 3: Set jurisdiction

Select the primary jurisdiction. PhronEdge maps the regulatory landscape:

• Required frameworks (GDPR, EU AI Act, NIST AI RMF, DORA, etc.)
• Required controls (up to 30 controls across 6 framework families)
• Transfer restrictions for cross-border data flows
• Risk score based on agent capabilities and data access

Step 4: Review intelligence

Before signing, the console shows:

• Risk score: 0-100 based on agent capabilities and jurisdiction
• Required controls: mapped from the regulatory intelligence engine
• Transfer corridors: cross-border data flow analysis
• Gaps: what controls are missing

Step 5: Sign

Click Sign Policy. The system:

1. 1.Generates a constitutional credential for each agent
2. 2.Signs each credential with ECDSA P-256
3. 3.Anchors the policy hash to an immutable chain
4. 4.Returns the credentials for your agents to use

After signing, the SDK automatically fetches the credential. No manual configuration.

Amendments

To change a signed policy:

1. 1.Click Amend on the active policy
2. 2.Modify agents, tools, or jurisdiction
3. 3.Review the diff (what changed from the previous version)
4. 4.Sign the amendment

Amendments create a new version linked to the previous policy hash. The full history is preserved and verifiable.

Observer

Real-time governance monitoring. Every event appears within seconds.

Event feed

Shows every governed action:

• TOOL_CALL_ALLOWED - tool call passed all checkpoints
• TOOL_CALL_BLOCKED - tool call blocked at a checkpoint
• PII_INPUT_DETECTED - personal data found in input
• PROMPT_INJECTION_DETECTED - injection attempt detected
• CREDENTIAL_VALIDATED - credential signature verified
• CREDENTIAL_REJECTED - invalid or tampered credential
• VAULT_TAMPER_DETECTED - credential modification detected
• VAULT_CREDENTIAL_RESTORED - original credential restored from vault
• AGENT_QUARANTINED - agent quarantined
• AGENT_REINSTATED - agent reinstated
• KILL_SWITCH_ACTIVATED - agent killed

Controls

• Kill switch: immediately revoke an agent. Takes effect globally in under 5 seconds.
• Quarantine: block all tool calls pending review. The agent stays alive but cannot execute.
• Reinstate: restore a quarantined agent.

Metrics

• Allow rate (percentage of calls that passed governance)
• Total allowed / blocked counts
• Tamper detection count
• System status (Observer, Judge, Enforcer, Brain, Anchor)

Audit Log

Every governance decision is anchored to a SHA-256 hash chain. Tamper one event and the chain breaks.

Features

• Full event history with cryptographic hashes
• Filter by agent, severity, event type, checkpoint
• Hash chain verification (one click to check integrity)
• Export as JSON or CSV
• Shareable links for external auditors (time-limited, token-based)

Chain verification

If the chain is intact, you see a green status. If any event has been modified, you see:

CHAIN BROKEN - CONTACT CISO IMMEDIATELY

This proves to any regulator that the audit trail has not been tampered with since the governance decision was made.

Sharing with auditors

1. 1.Click Share in the Audit Log
2. 2.Enter the auditor's email
3. 3.Set expiration (default 30 days)
4. 4.PhronEdge generates a secure link
5. 5.The auditor views the filtered audit log without needing an account

API Keys

Manage API keys for your agents.

• Keys start with pe_live_ (production)
• Shown in full only once on creation. After that, only the prefix is visible.
• Assign keys to specific agents or leave global
• Revoke instantly. The key is invalidated globally in under 5 seconds.

Usage

Monitor your plan consumption:

• Governed calls: calls this month vs your plan limit
• Agents: active agents vs plan limit
• API keys: active keys vs plan limit
• Team members: members vs plan limit
• Retention: how long events are stored

When you approach a limit, the console shows an upgrade prompt.

Settings

General

• Organization name
• Industry
• Primary jurisdiction

Team and Permissions

Invite team members and assign roles:

Notifications

Connect alert channels:

• Slack: paste your webhook URL and channel name
• PagerDuty: paste your routing key
• Microsoft Teams: paste your incoming webhook URL
• Email: alerts sent to team members based on their role

Configure alert rules:

• Choose which events trigger alerts (blocks, tamper, PII, injection, kill switch)
• Set severity filters
• Set cooldown periods (avoid alert fatigue)
• Route to specific channels

Integration

• Webhook URL: receive all governance events via HTTP POST
• API key: manage your SDK key
• Anchor network: where signed artifacts are anchored
• Anchor frequency: how often events are batch-anchored

Plans

Upgrade from the Plans page or Settings. Payment is handled securely. Plan changes take effect immediately.