zin-adk 0.1.0

Zin Agent Development Kit — build, deploy, and govern autonomous cybersecurity agents

Zin ADK — Agent Development Kit for Cybersecurity

Build, deploy, and govern autonomous cybersecurity agents.

ZIA provides a unified platform for building cybersecurity agents with bounded autonomy, policy enforcement, and full auditability — out of the box.

---

Install

pip install zin-adk

# With REST API support
pip install "zin-adk[api]"

# With graph database support (Python ≤ 3.13)
pip install "zin-adk[graph]"

Quick Start

# Scaffold a new agent
zia init --name "Vuln Triage Agent" --domain appsec --out ./agents

# Implement agents/vuln_triage_agent.py → execute()

# Validate
zia validate agents/vuln-triage-agent.yaml

# Run
zia run agents/vuln-triage-agent.yaml --tenant acme

# Or start the REST API
zia serve --host 0.0.0.0 --port 8000

What's included

Component	What it does
US-ADSL	YAML DSL for declarative agent definitions — schema-validated
Agent Runtime	Lifecycle management (pre_run → execute → post_run)
Policy Engine	6-rule in-process guardrails enforced on every tool call
Tool Substrate	@zia_tool decorator — policy-aware, auditable tool execution
Agent Registry	@register_agent — domain-based discovery and dispatch
SIF Graph	Time-aware security knowledge graph (Kuzu, namespace-isolated per tenant)
Multi-tenancy	Tenant isolation from day one — no leakage possible
REST API	FastAPI layer for portal and SIEM integration
CLI	zia init · zia validate · zia run · zia serve

Supported Agent Domains

Domain	Use case
risk_quant	Continuous risk scoring, FAIR-like modelling
supply_chain	Vendor posture, SBOM risk, 3rd-party exposure
red_team	Autonomous red teaming, attack path discovery
appsec	SAST, DAST, IaC scanning, dependency analysis
ai_security	Prompt risk, model exposure, LLM guardrail validation
compliance	Control monitoring, evidence collection, drift detection

Build an agent in 3 steps

1. Define it

# agents/my-agent.yaml
agent:
  id: my-agent
  domain: appsec
  version: "1.0.0"
intent:
  goal: "Triage open vulnerabilities by severity"
reasoning:
  mode: rule_based
  autonomy_level: bounded
capabilities:
  tools: [list_vulnerabilities]
boundaries:
  allowed_actions: [agent_execute, list_vulnerabilities]
safety:
  sandbox_profile: standard
  audit_level: standard

2. Implement it

from zia.core.runtime.agent import AgentContext, AgentResult, BaseAgent
from zia.core.runtime.registry import register_agent
from zia.core.tools.substrate import ToolExecutor
import zia.core.tools.builtins as tools

@register_agent(domain="appsec")
class MyAgent(BaseAgent):
    def execute(self, context: AgentContext) -> AgentResult:
        vulns = ToolExecutor.call(tools.list_vulnerabilities, context=context)
        critical = [v for v in vulns if v.get("severity") == "critical"]
        return AgentResult.ok(context, output={"critical_count": len(critical)})

3. Run it

zia run agents/my-agent.yaml --tenant acme

REST API

zia serve --host 0.0.0.0 --port 8000

# Create tenant
curl -X POST http://localhost:8000/tenants \
  -H "X-API-Key: your-key" \
  -d '{"tenant_id": "acme", "name": "Acme Corp"}'

# Trigger agent run
curl -X POST http://localhost:8000/tenants/acme/runs \
  -H "X-API-Key: your-key" \
  -d '{"domain": "risk_quant", "environment": "staging"}'

# Interactive API docs
open http://localhost:8000/docs

Documentation

Full docs: zeron.one/zia/docs

---

Zeron Security © 2026 — zeron.one