ClickHouse Pydantic Config

<!-- ADR: 2025-12-09-clickhouse-pydantic-config-skill -->

Generate DBeaver database client configurations from Pydantic v2 models using mise [env] as Single Source of Truth (SSoT).

Critical Design Principle: Semi-Prescriptive Adaptation

This skill is NOT a rigid template. It provides a SSoT pattern that MUST be adapted to each repository's structure and local database situation.

Why This Matters

Each repository has unique:

• Directory layouts (.dbeaver/ location may vary)
• Environment variable naming conventions
• Existing connection management patterns
• Local vs cloud database mix

The SSoT principle is the constant; the implementation details are the variables.

Quick Start

# Generate local connection config
mise run db-client-generate

# Generate cloud connection config
mise run db-client:cloud

# Preview without writing
mise run db-client:dry-run

# Launch DBeaver
mise run dbeaver

Credential Prerequisites (Cloud Mode)

<!-- ADR: 2025-12-10-clickhouse-skill-documentation-gaps -->

Before using cloud mode, obtain credentials via the skill chain:

1. Create/retrieve user: Use clickhouse-cloud-management skill to create read-only users or retrieve existing credentials from 1Password
2. Store in .env: Add to .env file (gitignored):

CLICKHOUSE_USER_READONLY=your_user
CLICKHOUSE_PASSWORD_READONLY=your_password

1. Generate config: Run mise run db-client:cloud

Skill chain: clickhouse-cloud-management → .env → clickhouse-pydantic-config

mise [env] as Single Source of Truth

All configurable values live in .mise.toml:

[env]
CLICKHOUSE_NAME = "clickhouse-local"
CLICKHOUSE_MODE = "local"  # "local" or "cloud"
CLICKHOUSE_HOST = "localhost"
CLICKHOUSE_PORT = "8123"
CLICKHOUSE_DATABASE = "default"

Scripts read from os.environ.get() with backward-compatible defaults—works with or without mise installed.

Credential Handling by Mode

| Mode | Approach | Rationale | | --------- | --------------------------------------- | ----------------------------------------------- | | Local | Hardcode default user, empty password | Zero friction, no security concern | | Cloud | Pre-populate from .env | Read from environment, write to gitignored JSON |

Key principle: The generated data-sources.json is gitignored anyway. Pre-populating credentials trades zero security risk for maximum developer convenience.

Cloud Credentials Setup

# .env (gitignored)
CLICKHOUSE_USER_READONLY=readonly_user
CLICKHOUSE_PASSWORD_READONLY=your-secret-password

Repository Adaptation Workflow

Pre-Implementation Discovery (Phase 0)

Before writing any code, the executor MUST:

# 1. Discover existing configuration patterns
fd -t f ".mise.toml" .
fd -t f ".env*" .
fd -t d ".dbeaver" .

# 2. Test ClickHouse connectivity (local)
clickhouse-client --host localhost --port 9000 --query "SELECT 1"

# 3. Check for existing connection configs
fd -t f "data-sources.json" .
fd -t f "dataSources.xml" .

Adaptation Decision Matrix

| Discovery Finding | Adaptation Action | | ---------------------------------- | ------------------------------------------------------ | | Existing .mise.toml at repo root | Extend existing [env] section, don't create new file | | Existing .dbeaver/ directory | Merge connections, preserve existing entries | | Non-standard CLICKHOUSE_* vars | Map to repository's naming convention | | Multiple databases (local + cloud) | Generate multiple connection entries | | No ClickHouse available | Warn and generate placeholder config |

Validation Checklist (Post-Generation)

The executor MUST verify:

• [ ] Generated JSON is valid (jq . .dbeaver/data-sources.json)
• [ ] DBeaver can import the config (launch and verify connection appears)
• [ ] mise tasks execute without error (mise run db-client-generate)
• [ ] .dbeaver/ added to .gitignore

Pydantic Model

The ClickHouseConnection model provides:

• Type-safe configuration with Pydantic v2 validation
• Computed fields for JDBC URL and connection ID
• Mode-aware defaults (cloud auto-enables SSL on port 8443)
• Environment loading via from_env() class method

See references/pydantic-model.md for complete model documentation.

DBeaver Format

DBeaver uses .dbeaver/data-sources.json with this structure:

{
  "folders": {},
  "connections": {
    "clickhouse-jdbc-{random-hex}": {
      "provider": "clickhouse",
      "driver": "com_clickhouse",
      "name": "Connection Name",
      "configuration": { ... }
    }
  }
}

Important: DBeaver does NOT support ${VAR} substitution—values must be pre-populated at generation time.

See references/dbeaver-format.md for complete format specification.

macOS Notes

1. DBeaver binary: Use /Applications/DBeaver.app/Contents/MacOS/dbeaver (NOT open -a)
2. Gitignore: Add .dbeaver/ to .gitignore

Related Skills

| Skill | Integration | | ------------------------------------------ | ----------------------------------- | | devops-tools:clickhouse-cloud-management | Credential retrieval for cloud mode | | quality-tools:clickhouse-architect | Schema design context | | itp:mise-configuration | SSoT environment variable patterns |

Python Driver Policy

For Python application code connecting to ClickHouse (not DBeaver), use clickhouse-connect (official HTTP driver). See clickhouse-architect for:

• Recommended code patterns
• Why NOT to use clickhouse-driver (community)
• Performance vs maintenance trade-offs

Additional Resources

| Reference | Content | | -------------------------------------------------------------- | ---------------------------- | | references/pydantic-model.md | Complete model documentation | | references/dbeaver-format.md | DBeaver JSON format spec |