Skip to content

configuration-reference.md

Latest commit

 

History

History
233 lines (166 loc) · 7.57 KB

configuration-reference.md

File metadata and controls

233 lines (166 loc) · 7.57 KB

Configuration Reference

All configuration is managed via backend/src/main/resources/application.properties. Sensitive values should use environment variable substitution.

Application Settings

Property Default Description
quarkus.application.name megabrain Application name
quarkus.application.version 1.0.0-SNAPSHOT Application version
quarkus.http.port 8080 HTTP server port
quarkus.http.cors true Enable CORS
quarkus.log.level INFO Root log level
quarkus.log.category."io.megabrain".level DEBUG MegaBrain log level

Search Configuration

Field Boosts

Control relevance ranking by boosting matches in specific fields.

Property Default Description
megabrain.search.boost.entity-name 3.0 Boost for entity name matches (class/method names)
megabrain.search.boost.doc-summary 2.0 Boost for documentation summary matches
megabrain.search.boost.content 1.0 Boost for code content matches

Hybrid Ranking Weights

Control the balance between keyword and vector search. Weights must sum to 1.0.

Property Default Description
megabrain.search.hybrid.keyword-weight 0.6 Weight for Lucene keyword search scores
megabrain.search.hybrid.vector-weight 0.4 Weight for vector similarity scores

Transitive Search

Configure depth limits for graph-based transitive searches.

Property Default Description
megabrain.search.transitive.default-depth 5 Default traversal depth when depth param not specified
megabrain.search.transitive.max-depth 10 Maximum allowed traversal depth (enforced on API)

Index Configuration

Property Default Description
megabrain.index.directory ./data/index Directory for Lucene index files
megabrain.index.batch.size 1000 Batch size for bulk indexing operations

Vector Search Configuration

Property Default Description
megabrain.vector.ef-search 40 HNSW ef_search parameter (higher = more accurate, slower)

Grammar Configuration

Property Default Description
megabrain.grammar.cache.directory ~/.megabrain/grammars Local cache directory for Tree-sitter grammar binaries

LLM Provider Configuration

Provider Selection

Property Default Description
megabrain.llm.provider ollama Active LLM provider: ollama, openai, or anthropic

Ollama (Free/Local)

Ollama inference uses only the configured base URL. When the endpoint is local (e.g. http://localhost:11434), no internet connectivity is required at runtime. Pull models beforehand with ollama pull <model> for fully offline operation (AC3).

Property Default Description
megabrain.llm.ollama.base-url http://localhost:11434 Ollama API endpoint
megabrain.llm.ollama.model codellama Model name (codellama, mistral, llama2, phi, etc.)
megabrain.llm.ollama.timeout-seconds 60 Request timeout in seconds

OpenAI (Paid Cloud)

Property Default Description
megabrain.llm.openai.api-key - OpenAI API key. Use ${OPENAI_API_KEY} for env var.
megabrain.llm.openai.chat-model.model-name gpt-3.5-turbo Model: gpt-3.5-turbo, gpt-4, gpt-4-turbo

Anthropic Claude (Paid Cloud)

Property Default Description
megabrain.llm.anthropic.api-key - Anthropic API key. Use ${ANTHROPIC_API_KEY} for env var.
megabrain.llm.anthropic.chat-model.model-name claude-3-sonnet-20240229 Model: claude-3-sonnet, claude-3-opus

Database Configuration

PostgreSQL (for vector search)

Property Default Description
quarkus.datasource.db-kind postgresql Database type
quarkus.datasource.username - Database username
quarkus.datasource.password - Database password. Use ${POSTGRES_PASSWORD}.
quarkus.datasource.jdbc.url - JDBC connection URL
quarkus.datasource.jdbc false Set to true when database is configured

Flyway Migrations

Property Default Description
quarkus.flyway.migrate-at-start false Run migrations on startup
quarkus.flyway.baseline-on-migrate false Baseline existing database

Neo4j (for graph database)

Property Default Description
megabrain.neo4j.uri - Neo4j Bolt URI (e.g., bolt://localhost:7687). When unset, graph queries return empty.
megabrain.neo4j.username neo4j Neo4j username
megabrain.neo4j.password - Neo4j password. Use ${NEO4J_PASSWORD}.

Source Control Configuration

GitHub

Property Default Description
github-api/mp-rest/url https://api.github.com GitHub API base URL
megabrain.github.token - GitHub PAT or App token. Use ${GITHUB_TOKEN}.

GitLab

Property Default Description
megabrain.gitlab.api-url https://gitlab.com GitLab API base URL (supports self-hosted)
megabrain.gitlab.token - GitLab access token. Use ${GITLAB_TOKEN}.
megabrain.gitlab.ssl.trust-store - Path to custom trust store (for self-hosted SSL)
megabrain.gitlab.ssl.verify-ssl true Enable SSL verification

Bitbucket

Property Default Description
bitbucket-cloud-api/mp-rest/url https://api.bitbucket.org Bitbucket Cloud API URL
bitbucket-server-api/mp-rest/url http://localhost:7990 Bitbucket Server/DC base URL (no /rest suffix)
megabrain.bitbucket.cloud.username - Bitbucket Cloud username
megabrain.bitbucket.cloud.app-password - Bitbucket Cloud app password
megabrain.bitbucket.server.username - Bitbucket Server username
megabrain.bitbucket.server.token - Bitbucket Server PAT

Frontend Configuration

Development Environment

Edit frontend/src/environments/environment.ts:

export const environment = {
  production: false,
  apiUrl: 'http://localhost:8080/api/v1'
};

Production Environment

Edit frontend/src/environments/environment.prod.ts:

export const environment = {
  production: true,
  apiUrl: '/api/v1'
};

API Proxy (Development)

The frontend/proxy.conf.json proxies API requests to the backend:

{
  "/api": {
    "target": "http://localhost:8080",
    "secure": false,
    "changeOrigin": true,
    "logLevel": "debug"
  }
}

Environment Variables

For sensitive values, use environment variables:

export GITHUB_TOKEN=ghp_...
export GITLAB_TOKEN=glpat_...
export OPENAI_API_KEY=sk-...
export ANTHROPIC_API_KEY=sk-...
export POSTGRES_PASSWORD=your_password
export NEO4J_PASSWORD=your_password
export BITBUCKET_CLOUD_USERNAME=your_username
export BITBUCKET_CLOUD_APP_PASSWORD=your_app_password
export BITBUCKET_SERVER_URL=https://bitbucket.mycompany.com
export BITBUCKET_SERVER_TOKEN=your_pat

Reference in application.properties:

megabrain.github.token=${GITHUB_TOKEN}
megabrain.llm.openai.api-key=${OPENAI_API_KEY}