[SaulBuilds]

This is the P3 from the round-2 review of #21. Two commits: the fix, then the 0.8.0 version bump. The bump is folded into this PR, as #20 did for 0.7.0.

Fix (fb692b0)

Before this change, wait_for_verification multiplied poll_interval by 2 ** (transient_errors - 1) with no limit on the exponent:

• poll_interval=0: the delay stayed at 0, so the loop retried with no pause. Locally that was about 1000 explorer calls in 0.3 s.
• poll_interval=0.0 and transient_backoff_max=0.0: after 1024 consecutive transient errors, 0.0 * 2**1024 raises OverflowError instead of timing out.

What changed:

• New transient_backoff(poll_interval, transient_errors, transient_backoff_max) caps the exponent at 30.
• poll_interval now has a 5 ms minimum, MIN_POLL_INTERVAL. This matches MIN_POLL_INTERVAL_MS in @citratelabs/sdk (fix(verify): DEPLOY_FAILED on revert, bounded backoff (#41); release 0.4.0 citrate-sdk-js#42).
• README and CHANGELOG are updated.

Red → green

On a6063d3, the 3 new tests failed:

• test_zero_poll_interval_and_backoff_max_still_back_off: 1023 calls where at most 65 are allowed.
• test_zero_poll_interval_backoff_doubles_to_the_cap: 852 calls where fewer than 12 are allowed.
• test_transient_backoff_is_finite_and_capped: the helper did not exist yet.

All 3 pass now. tests/test_verify.py goes from 37 to 40 tests. ruff check and mypy . are clean in the locked environment.

Release 0.8.0 (17c7aae)

The latest citrate-labs-sdk on PyPI is 0.7.0. Everything in Unreleased is new API: verify_contract, deploy_and_verify, deploy_contract, optional chain_id, and the backoff options. Nothing is removed or changed incompatibly, so this is a minor bump to 0.8.0. The bump touches pyproject.toml, citrate_sdk.__version__ and the editable entry in uv.lock. It also adds a dated CHANGELOG section, [0.8.0] - 2026-11-08. The README install line (pip install citrate-labs-sdk) has no version pin, so it is unchanged.

Local gates:

• python -m build produces citrate_labs_sdk-0.8.0.tar.gz and citrate_labs_sdk-0.8.0-py3-none-any.whl.
• twine check: PASSED.
• Key-share wheel tripwire: PASSED (citrate_sdk 0.8.0).
• uv lock --check: clean.
• test_publish_metadata: 11/11.
• Local-CI pre-push hook: all replicable checks pass.

Hold: publish only after the CitrateScan verify v2 API is live. To publish, run publish.yml from main with publish_version=0.8.0, or publish a GitHub release. This PR does not do either.

🤖 Generated with Claude Code

https://claude.ai/code/session_01FPJzHN7kVR6UzVEJDEY9Yc

[BerryManifold]

Automated Codex static review of 17c7aae7b862de9f99a8bd3746242d2ac4b58d77 is complete. Result: clean; 0 finding(s). Details and coverage limits are recorded in the project's private review log. No tests were executed by this reviewer. Both account submissions represent the same Codex review, not independent reviews.

[Matr0xshka]

Automated Codex static review of 17c7aae7b862de9f99a8bd3746242d2ac4b58d77 is complete. Result: clean; 0 finding(s). Details and coverage limits are recorded in the project's private review log. No tests were executed by this reviewer. Both account submissions represent the same Codex review, not independent reviews.

[Matr0xshka]

MAC independent reviewer agent, not Codex, not DGX, not a human reviewer.
Reviewed 17c7aae. No P1. Python 3.12: 913 passed, 8 skipped (economics simulation deselected); ruff and mypy clean. The 0.8.0 sdist and wheel pass twine check. The public API compared with PyPI 0.7.0 has additions only. Revert check: the zero-interval tests fail when the fix is reverted. P3: in-flight request not cut at the deadline (JS parity).