name	release-config-management
description	Standardize release & configuration management across chains/environments: deployments/<chainId>.json, env/runtime secrets, ABI/schema versioning, idempotent migrations, and reproducible scripts. Use when touching deploy scripts, env vars, DB migrations, ABI/events, or production setup.

Release & Config Management (Multi-chain + Multi-env)

When to use

Any PR touching:
- deploy scripts or addresses
- env vars or runtime config
- ABI/events/schema
- dbmate migrations / Kysely types
- production or testnet rollout

Goal (Definition of Done)

✅ Addresses are persisted per chainId in deployments/<chainId>.json.
✅ Env vars follow .env.example and secrets are never committed.
✅ DB migrations are idempotent and applied via dbmate.
✅ ABI/event schema changes are versioned and coordinated (contracts ↔ indexer ↔ app).
✅ Release has a clear “what changed” + upgrade/deploy steps.

Canonical config layout

deployments/<chainId>.json includes:
- GAME_ADDRESS and dependencies
- deploy blockNumber
- git commit hash
- timestamp
.env.example contains keys only (no values).
Runtime secrets are injected at runtime (not build time).

ABI/schema versioning

On any event/ABI change:

regenerate ABI artifacts
update Ponder handlers
add DB migration (if schema changes)
update API/app types
replay/reorg sanity tests

Database migrations (dbmate)

Migrations must be:
- forward-only and idempotent
- reviewed for locks/index impact
After migrations:
- regenerate Kysely types (if used)

Deployment checklist (minimum)

Contract deploy succeeds in target chainId
Explorer verification performed (recommended)
Indexer points to correct RPC + GAME_ADDRESS
App reads chainId + addresses correctly
Smoke E2E passes (pay → commit → reveal → draw → claim)

Anti-patterns (forbidden)

❌ Hardcoding addresses in frontend/backend.
❌ Committing .env or secrets.
❌ Changing ABI/events without updating indexer + schema.
❌ “Hotfixing” prod env without recording what changed.

Testing checklist

Local deploy + smoke (anvil)
Fork tests for external integrations (when relevant)
E2E (Playwright) for wallet → checkout → tx

Release & Config Management (Multi-chain + Multi-env)

When to use

Any PR touching:

deploy scripts or addresses
env vars or runtime config
ABI/events/schema
dbmate migrations / Kysely types
production or testnet rollout

Goal (Definition of Done)

✅ Addresses are persisted per chainId in deployments/<chainId>.json.

✅ Env vars follow .env.example and secrets are never committed.

✅ DB migrations are idempotent and applied via dbmate.

✅ ABI/event schema changes are versioned and coordinated (contracts ↔ indexer ↔ app).

✅ Release has a clear “what changed” + upgrade/deploy steps.

Canonical config layout

deployments/<chainId>.json includes:

GAME_ADDRESS and dependencies
deploy blockNumber
git commit hash
timestamp

.env.example contains keys only (no values).

Runtime secrets are injected at runtime (not build time).

ABI/schema versioning

On any event/ABI change:

regenerate ABI artifacts

update Ponder handlers

add DB migration (if schema changes)

update API/app types

replay/reorg sanity tests

Database migrations (dbmate)

Migrations must be:

forward-only and idempotent
reviewed for locks/index impact

After migrations:

regenerate Kysely types (if used)

Deployment checklist (minimum)

Contract deploy succeeds in target chainId

Explorer verification performed (recommended)

Indexer points to correct RPC + GAME_ADDRESS

App reads chainId + addresses correctly

Smoke E2E passes (pay → commit → reveal → draw → claim)

Anti-patterns (forbidden)

❌ Hardcoding addresses in frontend/backend.

❌ Committing .env or secrets.

❌ Changing ABI/events without updating indexer + schema.

❌ “Hotfixing” prod env without recording what changed.

Testing checklist

Local deploy + smoke (anvil)

Fork tests for external integrations (when relevant)

E2E (Playwright) for wallet → checkout → tx

release-config-management