set stage api urls to stage-blue for us and eu

[admin-dmou-ps]

No description provided.

[claude]

Review — one blocker, otherwise clean.

Correctness / test coverage — tests/api_uri_test.py:136 was missed, suite is red

test_each_shortcut_resolves (line 54) and test_stage_shortcut (line 127) were updated, but the per-shortcut EU test still asserts the old host:

    def test_stage_eu_shortcut(self):
        assert self._resolved_base("--stage-eu") == "https://api.stage-eu-v3.polyswarm.network/v3"

That fails against the new API_URI_SHORTCUTS["stage_eu"] = https://api.stage-eu-blue.polyswarm.network/v3. Fix to stage-eu-blue to match line 54.

Everything else checks out:

• Base is develop per the AGENTS.md gitflow rule; no pyproject.toml version bump, correct for a feature PR.
• Help strings for --stage / --stage-eu and the module docstring example were all updated in lockstep with the constant — no stale copies of the old hosts remain anywhere outside the one test line above.
• No spec drift: specs/01-architecture.md documents the shortcut names and the resolution precedence, not the concrete stage hosts, so nothing there needs editing. PROD_API_URI and the resolution logic are untouched.
• No downstream contract change — API_URI_SHORTCUTS values shift, but no signatures, exception classes, or SDK-boundary surface move.

[claude]

Reviewed against AGENTS.md and specs/01-architecture.md. The change itself is correct and complete: both API_URI_SHORTCUTS entries, both --help strings, and the module docstring example are updated in lockstep, and the tests cover both the constant-equality guard (test_each_shortcut_resolves) and the end-to-end resolved base for --stage / --stage-eu. No other references to the old hosts anywhere in the repo (no README, spec, or cassette mentions). specs/01-architecture.md describes the resolution mechanism and only hardcodes PROD_API_URI, so no spec update is owed here. Base is develop and no pyproject.toml version bump — both correct per the gitflow rules.

Two things:

1. Commit messages are missing conventional prefixes. AGENTS.md → "Conventional commit prefixes (feat:, fix:, refactor:, chore:, docs:, test:)". Both commits (set stage api urls to stage-blue for us and eu, update eu shortcut test) are bare. Suggest chore: point stage/stage-eu shortcuts at stage-blue ingress on a reword/squash. The PR body is also empty — a one-liner on why the ingress moved would help whoever bisects this in six months.

2. Deploy-slot coupling in a user-facing shortcut (design question, not a blocker). stage-blue / stage-eu-blue are blue/green slot hostnames, whereas the neighbouring entries (prod, prod-eu-v3) are slot-neutral. If stage ever cuts over to green, --stage and --stage-eu silently point at the idle stack, and the failure mode is confusing (stale-looking or 5xx responses, not a clear error). If a slot-neutral alias exists (e.g. an api.stage-v3 CNAME retargeted at whichever slot is live), that is the more durable thing to encode. If the -blue hosts are deliberately the long-lived stage endpoints now, worth saying so in the PR body so the next person does not 'fix"' the asymmetry back.

[claude]

Reviewed against AGENTS.md + specs/01-architecture.md. The URL swap itself is consistent: API_URI_SHORTCUTS, both --stage/--stage-eu help strings, the module docstring comment, and all three test assertions move together, and no spec hardcodes the staging hosts (specs/01-architecture.md only names the flags and the precedence rules), so no spec update is needed.

Two issues:

1. Version bump does not belong in this PR (AGENTS.md, Gitflow).
pyproject.toml 4.2.0 → 4.2.1, [tool.bumpversion] current_version, and src/polyswarm/__init__.py are all bumped here. AGENTS.md:

PyPI release happens automatically when pyproject.toml's version changes on master. Don't bump the version inside a feature PR unless the maintainer specifically asks — version bumps belong to the develop → master step.

Concretely: carrying the bump on the feature branch means the next develop → master PR fires a PyPI release for 4.2.1 covering whatever else has accumulated on develop, not just this change — the release decision gets made implicitly by this merge rather than by the maintainer at release time. Please drop the three version edits unless a maintainer asked for them. (Base branch is correct — develop.)

2. Is stage-blue / stage-eu-blue a stable alias or a deploy slot?
If those hosts are blue/green deploy slots, pinning --stage to the blue slot means the shortcut silently points at the wrong environment after every swap, and tests/api_uri_test.py:48 (# guard against the constant drifting from the verified ingress hosts) will hard-fail the suite on each swap rather than just being stale. If a slot-independent alias (e.g. the previous api.stage-v3… name) still resolves to the active stage ingress, that is the better target. If blue is genuinely the permanent stage ingress now, say so in the PR body — the PR description is currently empty, so there is no record of why the hosts moved.

No correctness or coverage gaps otherwise; the shortcut resolution logic is untouched and the parametrised host assertions cover both new values.