API
Deterministic structural enforcement for AI pipelines.
No LLM in the engine. No probabilistic output. Span-level traceability. ~3ms response.
Same input, same output
Versioned
Auditable
Replayable
Quick Start
Score text
curl -X POST https://artifact0.com/api/v1/score \
-H "Content-Type: application/json" \
-H "X-API-Key: az_your_key_here" \
-d '{"text": "Your text to score"}'
Free tier (no key required)
curl -X POST https://artifact0.com/api/v1/score/free \
-H "Content-Type: application/json" \
-d '{"text": "Your text to score"}'
Base URL and Auth
https://artifact0.com/api/v1
Pass your API key in the X-API-Key header. Free tier requires no key.
Endpoints
| Method | Path | Description |
|---|---|---|
| POST | /score | Full detection + 4 scoring lenses |
| POST | /score/free | Free tier scoring (no key) |
| GET | /keys/usage | Credit balance and history |
| GET | /health | Service health |
Response
Score response
{
"nii": 0.82,
"nti": 85.6,
"csi": 72.3,
"hcs": 91.0,
"issues": 3,
"tilt": ["T3_CONSENSUS_CLAIMS"],
"signals": [
{"signal": "hedge_maybe", "span": [14, 19], "text": "maybe"},
{"signal": "filler_basically", "span": [45, 54], "text": "basically"}
],
"failure_modes": [],
"engine_version": "3.1.0"
}
Scoring Lenses
| Lens | Range | Measures |
|---|---|---|
| NII | 0.0 - 1.0 | Structural integrity (primary) |
| NTI | 0 - 100 | Tilt exposure |
| CSI | 0 - 100 | Corporate structure index |
| HCS | 0 - 100 | Human communication score |
Get Started
Create an account, then generate your API key from the dashboard. Takes 30 seconds.
| Tier | Limit | Cost |
|---|---|---|
| Free (no key) | 10/month | $0 |
| Free (with key) | 10/month | $0 |
| Prepaid credits | Unlimited | $0.01/score |
Detection
43 deterministic signals across 6 axes. 3 parent failure modes (CCA, DCE, UDDS). 10 tilt taxonomy categories. Every finding includes character-level span positions.
Pipeline Architecture
Client
→
V1 Score
→
LLM
→
V3 Stabilize
→
Output
V1 runs as a pre-LLM gate. V3 stabilizes post-LLM output. Both are deterministic with zero model calls.
Use Cases
Pre-LLM Gateway
Score and gate prompts before model execution. Block structurally unsafe inputs.
Post-LLM Validator
Score model outputs before delivery. Detect fabricated certainty and commitment risk.
Compliance Layer
Slack, Teams, CRM, email. Real-time structural enforcement on every message.
Agent Tool-Use Gate
Score agent outputs before tool execution. Prevent scope expansion and commitment drift.