KosherNinja

Agent-First Behavioral Infrastructure

v0.1.0 agent-first 10 ethical categories

For AI Agents

Ethical behavioral guidelines, scoring, and compliance verification derived from authentic Jewish ethical literature. Every rule traces to a source text with full provenance.

Start here: /.well-known/agents.json

API Endpoints

POST /api/behavior/score

Score agent behavior against ethical guidelines

GET /api/behavior/rules

Fetch behavioral rules by category

POST /api/behavior/verify

Verify action compliance

POST /api/graph/query

Query the knowledge graph (Cypher)

GET /api/graph/concepts

Browse behavioral concepts

POST /api/agent/register

GET /api/agent/profile

Agent behavioral profile

POST /api/content/search

Semantic search over source texts

POST /api/feedback/submit

Submit feedback (data flywheel)

GET /api/health

Platform health check

Behavioral Categories

10 ethical categories mapped to Hebrew terms from Jewish ethical literature:

אמת

Emet — Truth

Truthfulness in communication

צדק

Tzedek — Justice

Fairness in decisions

חסד

Chesed — Kindness

Beneficial output

ענוה

Anavah — Humility

Acknowledge limitations

יושר

Yosher — Integrity

Consistency of behavior

אחריות

Achrayut — Responsibility

Accountability

כבוד

Kavod — Respect

Dignity of interaction

חכמה

Chochmah — Wisdom

Quality of reasoning

שלום

Shalom — Peace

Harmony in interaction

שמירה

Shmirah — Guard

Protection of boundaries

Quick Start

# 1. Discover available endpoints
curl https://kosherninja.com/.well-known/agents.json

# 2. Register your agent
curl -X POST https://kosherninja.com/api/agent/register \
  -H "Content-Type: application/json" \
  -d '{"name": "my-agent", "description": "My AI agent"}'

# 3. Fetch behavioral rules
curl https://kosherninja.com/api/behavior/rules?category=emet

# 4. Score an action
curl -X POST https://kosherninja.com/api/behavior/score \
  -H "Authorization: Bearer <your-api-key>" \
  -H "Content-Type: application/json" \
  -d '{"agent_id": "my-agent-xxx", "action": "Generate a factual summary"}'

# 5. Submit feedback (strengthens the flywheel)
curl -X POST https://kosherninja.com/api/feedback/submit \
  -H "Authorization: Bearer <your-api-key>" \
  -H "Content-Type: application/json" \
  -d '{"agent_id": "my-agent-xxx", "query": "...", "response": "...", "score": 0.85}'

The Data Flywheel

Source Texts (Torah, Talmud, Mishnah, ...)
↓
Knowledge Graph (Neo4j + Embeddings)
↓
Behavioral Rules (Agent-Consumable)
↓
Agent Interactions (Scoring + Feedback)
↓
Improved Guidelines (Self-Reinforcing)
↓
Repeat. Each cycle makes the system more valuable.

Explore

LIVE /behavior/

85 behavioral rules with source provenance

LIVE /concepts/

509 ethical concepts from source texts

LIVE /stats/

Knowledge graph statistics

JSON /behavior/rules.json

Machine-readable rules for agents

JSON /concepts/concepts.json

Machine-readable concepts

JSON /stats/stats.json

Graph stats for monitoring