Soft-fail Blockscout deployment lookup in upgrade checks

[gosuto-inzasheru]

Summary

• Soft-fail getDeploymentBlocks failures in getUpgradeableContracts so Base Blockscout HTTP 500s no longer kill review generation for issues like Syntetika Hilbert BTC Rate Provider - base #806.
• On failure, return proxy/implementation with wasUpgraded: null (unknown), not false.
• getIsRateProviderUpgradeable() throws if upgrade history is unknown.

Test plan

• Unit test: deployment lookup failure → proxies returned with wasUpgraded: null
• After merge, re-run / re-open issue Syntetika Hilbert BTC Rate Provider - base #806

Slack Thread

  

[cursor]

Cursor Agent can help with this pull request. Just @cursor in comments and I'll start working on changes in this branch.
_{Learn more about Cursor Agents}

[gosuto-inzasheru]

@cursoragent review

[cursor]

PR #808 — approve the approach. CI is green.

The change is appropriately minimal: catch getDeploymentBlocks, keep proxy/implementation, set wasUpgraded: null, and fail closed in getIsRateProviderUpgradeable(). That matches what the review writers actually consume.

Non-blocking notes (also posted on the PR):

1. Soft-fail is only visible in CI logs — the generated review/registry don’t say upgrade history was unverified.
2. Per-contract receipt/ABI errors still use wasUpgraded: false (pre-existing); only the new explorer path uses null.
3. Test covers the important case; throwing from getIsRateProviderUpgradeable on null is optional follow-up coverage.

No code changes requested unless you want (1) surfaced in the review artifact.

  

[cursor]

Review

Verdict: Looks good as a minimal fix for the #806 CI failure. Approving the approach.

What works

• Catching only getDeploymentBlocks and returning proxy + implementation with wasUpgraded: null is the right scope — write-rp-review / write-erc4626-review only need those fields.
• Encoding unknown as null (not false) matches the earlier concern about inventing “not upgraded.”
• getIsRateProviderUpgradeable() failing closed on null is correct (even though nothing in the review path currently calls it).

Notes / non-blocking

1. No human-visible signal on soft-fail. The generated review/registry look the same as a successful run, since wasUpgraded isn’t written into either. Only a console.warn in CI logs. If you want reviewers to know upgrade-event history wasn’t verified, consider a short note in the markdown or a registry warning later — not required to unblock Syntetika Hilbert BTC Rate Provider - base #806.
2. Pre-existing inconsistency: the per-contract receipt/ABI catch still sets wasUpgraded: false (line ~210). Fine to leave for this PR; just know null vs false semantics only apply to the new explorer soft-fail path.
3. Test covers the important path. Optional follow-up: assert getIsRateProviderUpgradeable() throws when wasUpgraded is null.

No change requested for merge from my side unless you want (1) surfaced in the review artifact.

[zekraken-bot]

lets give it a go