Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.plyra.dev/llms.txt

Use this file to discover all available pages before exploring further.

The hosted Plyra instance runs on Azure Container Apps in the Central India region. To run your own: 
# 1. Build and push the image
docker build -t your-registry/plyra-memory-server:latest .
docker push your-registry/plyra-memory-server:latest

# 2. Create the container app
az containerapp create \
  --name plyra-memory-server \
  --resource-group your-rg \
  --image your-registry/plyra-memory-server:latest \
  --target-port 7700 \
  --ingress external \
  --secrets \
    admin-key=your_admin_key \
    groq-key=your_groq_key \
  --env-vars \
    PLYRA_ADMIN_API_KEY=secretref:admin-key \
    GROQ_API_KEY=secretref:groq-key

​
Environment variables

VariableRequiredDescription
PLYRA_ADMIN_API_KEY✅Admin key for key management routes
GROQ_API_KEY—LLM extraction (Groq is preferred)
ANTHROPIC_API_KEY—Fallback LLM provider
OPENAI_API_KEY—Fallback LLM provider
PLYRA_DATABASE_URL—Postgres URL for multi-node deployments
PLYRA_SERVER_PORT—Default: 7700
PLYRA_LOG_LEVEL—Default: INFO

​
LLM provider priority

The server uses LLMs to extract structured facts from free-text memories. Provider priority:
  1. Groq (GROQ_API_KEY) — preferred, fastest
  2. Anthropic (ANTHROPIC_API_KEY) — fallback
  3. OpenAI (OPENAI_API_KEY) — fallback
  4. Regex — no LLM required, reduced extraction quality

​
Postgres (multi-node)

For deployments with multiple server instances, use Postgres instead of SQLite: 
export PLYRA_DATABASE_URL=postgresql://user:pass@host:5432/plyra_memory

​
Health check

curl https://your-server/health
# → {"status": "ok", "version": "0.3.0"}
Always set PLYRA_ADMIN_API_KEY to a strong random value before deployment. The default is insecure and must be changed.