Problem
With npm run lint:registry already validating structure, there's an open question of whether blueprints actually produce correct translations against real-world event payloads over time — especially since contracts can be upgraded (changing event shapes) without any corresponding blueprint update, silently producing wrong-but-plausible-looking sentences.
Proposed work
Build a fixtures directory of real captured testnet/mainnet raw events (paired with their expected translated output) per blueprint.
Add a test runner that feeds each fixture through the live blueprint and asserts exact output match, run in CI alongside lint:registry.
When a contract's WASM hash changes (tying into verification ideas from contract upgrades), flag any blueprint whose fixtures no longer match as needs-review rather than letting it fail silently in production.
Acceptance criteria
At least 5 existing blueprints have committed fixture-based regression tests.
CI fails clearly when a blueprint's output drifts from its expected fixture, distinct from a generic test failure.
Process for contributors to add fixtures for their own new blueprints is documented in CONTRIBUTING.md.
Problem
With npm run lint:registry already validating structure, there's an open question of whether blueprints actually produce correct translations against real-world event payloads over time — especially since contracts can be upgraded (changing event shapes) without any corresponding blueprint update, silently producing wrong-but-plausible-looking sentences.
Proposed work
Build a fixtures directory of real captured testnet/mainnet raw events (paired with their expected translated output) per blueprint.
Add a test runner that feeds each fixture through the live blueprint and asserts exact output match, run in CI alongside lint:registry.
When a contract's WASM hash changes (tying into verification ideas from contract upgrades), flag any blueprint whose fixtures no longer match as needs-review rather than letting it fail silently in production.
Acceptance criteria
At least 5 existing blueprints have committed fixture-based regression tests.
CI fails clearly when a blueprint's output drifts from its expected fixture, distinct from a generic test failure.
Process for contributors to add fixtures for their own new blueprints is documented in CONTRIBUTING.md.