[arosasg]

Problem

AGENTS.md, CLAUDE.md and docs/DESIGN.md section 13 still said nothing may be tagged or published before 2026-11-01 without Alex's explicit approval. Alex approved 1.0.0 on 2026-10-29 (now published) and asked that every release publish straight from the repository.

Solution

• The agent rule now says merging the release-please pull request is the release approval; nobody tags or publishes by hand outside that flow, and no registry token is stored where trusted publishing works.
• docs/DESIGN.md section 13 records the 1.0.0 release, the publish path per registry and the workflow_dispatch hand-off; the file stays byte-identical in all seven SDK repositories (sha1 9aec8cc2).

Verification

• shasum docs/DESIGN.md is identical in sdk-python, sdk-node, sdk-go, sdk-java, sdk-php, sdk-ruby and sdk-dotnet.
• Documentation only; no workflow or code changes.

🤖 Generated with Claude Code

[arosasg]

Codex review (codex CLI, account codex-7, model gpt-5.6-sol)

---

💡 Code review

Reviewed commit: d8a1899

Align the release workflow with merge-as-approval

docs/DESIGN.md:L252 - view

When the next release-please PR is merged, the publish job still targets the protected rubygems environment whose required reviewer is mandated by CONTRIBUTING.md, so publishing pauses for an additional manual approval despite this new contract saying the merge itself is approval. The workflow comments and CONTRIBUTING.md also retain the superseded pre-November approval policy; update the release configuration and operational documentation so releases actually publish under the newly documented process.

Verdict: REQUEST CHANGES

[arosasg]

Codex review (codex CLI, account codex-7, model gpt-5.6-sol)

---

💡 Code review

Reviewed commit: 84a85b9

Remove the obsolete publish reviewer instruction

CONTRIBUTING.md:L76-L78 - view

When a maintainer follows the updated release procedure, line 78 says merging the release PR is the approval while line 76 still says publishing waits for a rubygems environment reviewer; because the environment now explicitly has no required reviewers, this leaves the operational instructions contradictory and implies a nonexistent manual gate. Update the workflow description to say publishing proceeds automatically after the release PR is merged.

Update the completed RubyGems setup instructions

CONTRIBUTING.md:L81-L85 - view

For maintainers consulting the section now labeled as completed for 1.0.0, step 2 still claims didit-sdk does not exist and directs them to create a pending trusted publisher; after the documented 1.0.0 release the gem exists, so following this instruction targets the wrong setup flow. Record the publisher as already configured, or describe the normal existing-gem trusted-publisher configuration instead.

Verdict: APPROVE