release: v7.1.0 — OpenAPI 3.25.0 additive sync

[TexasCoding]

Summary

Prepare v7.1.0 (minor) after the #476 OpenAPI 3.25.0 reconcile.

Version bump: minor (7.0.0 → 7.1.0)

Kind	Why not
Major	No removals, renames, or incompatible method signature changes
Patch	New public type (MarginMarketSchedule) + new required response field — new capability, not a bugfix
Minor	Additive public surface + upstream spec pin 3.24.0 → 3.25.0

Changes

• Bump pyproject.toml / kalshi.__version__ → 7.1.0
• CHANGELOG.md section for 7.1.0 (release-workflow body source)
• Docs: README, docs/index.md, docs/perps.md (schedule field), docs/migration.md (v7.0→v7.1), Claude.md OpenAPI pin, ROADMAP.md shipped entry
• Refresh stale perps/Klear operation counts (35 REST / 11 SCM)

Release steps (after merge)

git checkout main && git pull
git tag v7.1.0
git push origin v7.1.0

release.yml will build, publish to PyPI (OIDC / pypi env), and create the GitHub Release from the CHANGELOG section.

Test plan

• Version strings match across pyproject + __version__ + CHANGELOG heading
• CI green
• Merge → tag v7.1.0 → confirm release workflow + PyPI

[claude]

Review

This is a clean, docs/version-bump-only release-prep PR following the #476 spec-drift reconcile — no functional code changes. Reviewed the diff against CLAUDE.md's release conventions and cross-checked the claims against the repo.

Summary

• Bumps pyproject.toml / kalshi.__version__ 7.0.0 → 7.1.0, adds a matching CHANGELOG.md section, and refreshes operation-count docs (README, docs/index.md, docs/perps.md, docs/migration.md, CLAUDE.md, ROADMAP.md).
• Version classification (minor, not patch/major) is correctly justified: MarginMarketSchedule is additive, and there are no removals/renames.

Verified

• Version strings are consistent across pyproject.toml, kalshi/__init__.py, and the CHANGELOG.md heading (all 7.1.0).
• Operation counts check out: traced PERPS_METHOD_ENDPOINT_MAP / PERPS_SCM_METHOD_ENDPOINT_MAP in tests/_contract_support.py — deduping pagination-helper (*_all) entries that share an endpoint with their non-paginated sibling yields exactly 35 unique perps REST operations and 11 unique Klear operations, matching the new README/docs numbers.
• The core REST count intentionally stays at 101 (spec now has 102 operations; POST /portfolio/intra_exchange_instance_transfer is recorded in _UNIMPLEMENTED_ENDPOINTS since it's upstream-marked "currently not available" and already covered by PerpsClient.transfers.transfer_instance()) — CLAUDE.md's wording reflects this correctly.
• No stale v3.24.0/old-count references were left behind in the files this PR touches.

Minor nit

• docs/migration.md: the new heading is ## v7.0 → v7.1.0, but every other entry in the file uses the bare major version for the "from" side (## v6 → v7.0.0, ## v5 → v6.0.0, ## v4 → v5.0.0, ...). For consistency this should probably be ## v7 → v7.1.0. Cosmetic only, doesn't block merge.

Risk

Low — no kalshi/ source or test changes, so nothing for mypy/drift tests to catch. The unchecked "CI green" and "merge → tag → confirm release" boxes in the test plan are expected for a still-open PR.

Nice work keeping the doc/version bump surgical and getting the op-counts actually right instead of just incrementing them.

[claude]

Review: release: v7.1.0 — OpenAPI 3.25.0 additive sync

Overview
Mechanical release-prep PR: version bump (7.0.0 to 7.1.0), CHANGELOG/ROADMAP/migration-guide entries, and refreshed operation counts in README/docs, following the MarginMarket.schedule addition and OpenAPI 3.25.0 pin that landed in the prior PR (476). No kalshi/ source changes beyond the version string in init.py / pyproject.toml. Low risk given the scope.

Code quality / consistency — looks good

• Version string is consistent across pyproject.toml, kalshi/init.py, and the CHANGELOG.md heading (satisfies the PR's own test-plan checkbox).
• The MarginMarketSchedule field, its exports (kalshi.perps, kalshi.perps.models), and the _UNIMPLEMENTED_ENDPOINTS entry for POST /portfolio/intra_exchange_instance_transfer referenced in the CHANGELOG/CLAUDE.md all check out against the current tree (kalshi/perps/models/markets.py, kalshi/perps/init.py, tests/test_contracts.py lines 1251-1259).
• CHANGELOG/migration.md/ROADMAP.md entries follow the established style of prior release entries (e.g. 7.0.0).
• ROADMAP.md also backfills a v7.0.0 Shipped bullet that appears to have been missing before — reasonable cleanup, but it is outside what the PR description says it does; worth a one-line mention in the PR body so reviewers are not surprised.

Possible issue — operation counts may still be off
The PR explicitly calls out refreshing stale perps/Klear operation counts (35 REST / 11 SCM). Counting distinct (http_method, path_template) pairs in tests/_contract_support.py's PERPS_METHOD_ENDPOINT_MAP and PERPS_SCM_METHOD_ENDPOINT_MAP (the source of truth the drift tests use) yields 39 and 13 respectively, not 35/11. These doc counts are not asserted anywhere in CI (no test compares len(PERPS_METHOD_ENDPOINT_MAP) against the README string), so a mismatch will not fail the build — it will just quietly drift again. Worth double-checking the exact numbers before merge, or better, deriving them programmatically from the map lengths rather than hand-counting, to avoid the same staleness recurring next release.

Test coverage
No functional code changed, so no new tests are needed here. The test-plan checklist (version-string match done, CI green / tag-and-release pending) is appropriate for a release PR.

Security
No concerns — docs and version metadata only.

Minor

• docs/migration.md correctly flags the practical breaking edge case (hand-constructed MarginMarket instances in tests now need schedule=) even though the PR is scoped as minor per semver — good, matches the precedent set for the earlier MarginPosition.subaccount required-field addition.

Overall: safe to merge once the perps/Klear operation counts are confirmed or fixed.