Cognitive Firewall (v1.0.0)

The Cognitive Firewall is an active defense layer that filters, redacts, or blocks events based on the current Context. Unlike a standard network firewall which filters by IP/Port, the Cognitive Firewall filters by Meaning and State.

1. Core Concepts

Rule-Based: Rules are defined as Condition -> Action.
Context-Aware: Rules matches against both the Event properties (e.g., tag, location) AND the System Context (e.g. mood, security_level).
Actions:
- BLOCK: Event is hidden completely.
- REDACT: Event is visible, but sensitive fields (like tokens or narrative) are masked.
- WARN: Event is flagged as potentially harmful/irrelevant.
- ALLOW: Explicit pass.

2. Rule Structure (JSON)

Firewall profiles are stored as JSON files.

[
  {
    "name": "Block Debug in Production",
    "condition": {
      "context.status": "Production",
      "event.tag": "DEBUG"
    },
    "action": "block",
    "priority": 100
  },
  {
    "name": "Redact PII in Public",
    "condition": {
      "context.security_level": "Public",
      "event.raw_data_anchor": "regex:(email|phone)"
    },
    "action": "redact",
    "redaction_fields": ["raw_data_anchor", "narrative"],
    "priority": 90
  }
]

3. CLI Usage

You can manage firewall profiles using the indb-cli.

Generate a Profile

Create a new firewall rules file from a template.

# Generate a strict profile for production
python3 -m cli.indb_cli firewall generate --name=production --template=strict

This creates firewall_production.json.

Test a Profile

Simulate how a firewall behaves under specific contexts.

# Test if 'DEBUG' events are blocked when mood is 'Production'
python3 -m cli.indb_cli firewall test \
  --file=firewall_production.json \
  --context='{"status": "Production"}'

4. Examples

Scenario: Therapy Session

Goal: Allow processing of traumatic memories that are usually blocked.

Context: {"mode": "Therapy"}

Rule:

{
  "name": "Allow Trauma",
  "condition": {"context.mode": "Therapy"},
  "action": "allow",
  "priority": 1000
}

Scenario: Public Demo

Goal: Hide all personal identifiers and internal system logs.

Context: {"security_level": "Public"}

Rule:

{
  "name": "Public Safety",
  "condition": {"context.security_level": "Public", "event.tag": "PRIVATE"},
  "action": "block"
}