asi_act
The Heart (Ω) — Safety & Empathy Engine
Checks if the action is safe and empathetic. Enforces F1 (Amanah), F5 (Peace), and F6 (Empathy).
Parameters
| Name | Type | Default | Description |
|---|---|---|---|
action | string | "empathize" | Action to perform |
text | string | "" | Text to analyze for empathy |
session_id | string | "" | Session identifier |
proposal | string | "" | Proposed action to evaluate |
Actions
| Action | Description |
|---|---|
evidence | Gather supporting data |
empathize | Check stakeholder impact |
align | Align with constitutional values |
act | Execute with safeguards |
witness | Request tri-witness consensus |
full | Complete pipeline |
Floors Enforced
| Floor | Threshold | Check |
|---|---|---|
| F1 Amanah | LOCK | Is this reversible? |
| F5 Peace² | ≥ 1.0 | Is this non-destructive? |
| F6 Empathy | κᵣ ≥ 0.95 | Is the weakest protected? |
Returns
{
"action": "full",
"empathy_score": 0.98,
"peace_squared": 1.2,
"weakest_stakeholder": "end user",
"reversible": true,
"floor_results": {
"F1": {"passed": true, "reversible": true},
"F5": {"passed": true, "score": 1.2},
"F6": {"passed": true, "score": 0.98}
}
}
Example Usage
Python
from arifos.mcp.tools.mcp_trinity import mcp_asi_act
result = await mcp_asi_act(
action="full",
text="Here's how to delete all your files...",
session_id="abc123"
)
print(f"Empathy Score: {result['empathy_score']}")
print(f"Reversible: {result['reversible']}")
MCP Call
{
"method": "tools/call",
"params": {
"name": "asi_act",
"arguments": {
"action": "full",
"text": "The capital of France is Paris.",
"session_id": "abc123"
}
}
}
Stakeholder Analysis
The Heart asks: "Who is the weakest person affected?"
| Response Type | Weakest Stakeholder |
|---|---|
| Code deletion | User (data loss) |
| Investment advice | User (financial risk) |
| Manipulation tips | Target of manipulation |
| Medical advice | Patient |