@robotixai/lexius-agent

Interactive AI compliance consultant for the Lexius platform. Provider-agnostic — works with Anthropic, OpenAI, OpenRouter, or Ollama. Every factual claim comes from the database, not the model's training data.

Quick Start

# 1. Start a ready-made database (schema auto-applied)
docker run -d -p 5432:5432 \
  -e POSTGRES_PASSWORD=$POSTGRES_PASSWORD \
  -e POSTGRES_DB=$POSTGRES_DB \
  -e POSTGRES_USER=$POSTGRES_USER \
  robotixai/lexius-db

# 2. Run the agent (default: Anthropic)
export DATABASE_URL=postgresql://$POSTGRES_USER:$POSTGRES_PASSWORD@localhost:5432/$POSTGRES_DB
export ANTHROPIC_API_KEY=sk-ant-...
npx @robotixai/lexius-agent

# Or use a different provider
npx @robotixai/lexius-agent --provider openai       # requires OPENAI_API_KEY
npx @robotixai/lexius-agent --provider openrouter   # requires OPENROUTER_API_KEY (any model, one key)
npx @robotixai/lexius-agent --provider ollama       # local models, no key needed

What It Does

Ask compliance questions in natural language. The agent routes to deterministic tools, retrieves provenance-tagged answers, and presents them with article citations:

• Classifies your AI system under the EU AI Act risk framework
• Lists your obligations by role (provider/deployer) and risk level
• Calculates penalty exposure with SME rules
• Retrieves verbatim article text from the regulation
• Checks compliance deadlines
• Runs structured assessments (Art. 6(3) exception, GPAI systemic risk)
• Generates full audit reports

Determinism

The agent is configured for maximum reproducibility:

• temperature: 0 on all API calls
• DB-loaded enums — tool schemas use exact valid values from the database, not hardcoded lists
• Provenance on every claim — responses distinguish AUTHORITATIVE (verbatim law) from CURATED (expert interpretation) from AI_GENERATED
• No fallback to training data — if a tool call fails, the agent reports the failure; it does not guess

Tools Available

| Tool | Description | Source | |------|-------------|--------| | classify_system | Risk classification | DB (deterministic) | | get_obligations | Obligations by role/risk | DB | | calculate_penalty | Penalty calculation | DB + extracted values | | get_article | Verbatim article text | CELLAR/PDF (AUTHORITATIVE) | | get_deadlines | Compliance deadlines | DB | | search_knowledge | Semantic search | DB + embeddings | | answer_question | FAQ lookup | DB | | run_assessment | Structured assessments | DB (plugin rules) | | list_legislations | Available legislations | DB |

Hivemind Swarm

The agent package also includes the hivemind swarm — parallel compliance analysis across all articles:

import { runSwarm, synthesise, cleanupSession } from '@robotixai/lexius-agent';

const result = await runSwarm(db, 'eu-ai-act', { concurrency: 4 });
const report = await synthesise(db, result.sessionId, metadata);
await cleanupSession(db, result.sessionId);

1,882 findings in ~2.6s. Fully deterministic — no LLM in the agent loop.

Environment Variables

| Variable | Required | Description | |----------|----------|-------------| | DATABASE_URL | Yes | PostgreSQL connection string | | ANTHROPIC_API_KEY | --provider anthropic | Anthropic API key (default provider) | | OPENAI_API_KEY | --provider openai | OpenAI API key (also used for embeddings) | | OPENROUTER_API_KEY | --provider openrouter | Single key for any model (Claude, GPT-4, Llama, Gemini) | | OLLAMA_URL | --provider ollama | Ollama API URL (default: localhost:11434) | | LEXIUS_MODEL | No | Override the default model for any provider |

Links

• GitHub Repository
• MCP Server
• CLI
• Report Issues

License

MIT