[unbalancedparentheses]

Smaller releases: 0.9 introspection and the attachable console (design approved), 0.10 unified start and the dynamic supervisor, 0.11 the admission package. Adds how each feature is implemented against its approved design (test plan as acceptance checklist, trace validation in the same PR, planted bugs, no-cost benchmark, final review), and widens the code-coverage report row to the whole crate over the simulated suites (report only).

[github-actions]

🤖 DeepSeek Code Review

This PR is documentation-only: it updates CHANGELOG.md and ROADMAP.md with release planning/process details and re-scopes upcoming work. There are no source, test, or dependency changes, so the requested code-review areas (correctness, concurrency, performance, Rust best practices, memory safety, error handling, actor lifecycle/message-passing/cancellation) do not apply.

No bugs or risks found in the diff.

One minor editorial note: the new CHANGELOG.md “ROADMAP…” bullet under Unreleased is a process/planning item rather than a user-facing behavior change. If the changelog is meant for end-user release notes, consider moving that to ROADMAP.md only or rewording it as a documentation highlight. Otherwise LGTM.

---

Automated review by DeepSeek · deepseek-v4-pro · custom prompt

[github-actions]

🤖 Gemini Code Review

The changes in ROADMAP.md and CHANGELOG.md are clear and well-structured. splitting 0.9 into focused releases (0.9 for introspection/console and 0.10 for unified start/dynamic supervision) and documenting the per-feature acceptance criteria and expanded coverage scope looks good. LGTM.

---

Automated review by Gemini · gemini-4-argon · custom prompt

[github-actions]

🤖 Grok Code Review

Documentation-only change. No actor lifecycle, concurrency, or memory-safety issues in this diff.

Changelog overclaims the coverage change. CHANGELOG.md says the coverage report was widened to the whole crate over the simulated suites. This diff only rewrites the “Ongoing” row in ROADMAP.md (around the old line 400 / new line 415). There is no cargo-llvm-cov config, CI workflow, or harness change. Either land that measurement change in this PR, or phrase the changelog as a roadmap policy update, not completed work.

Release renumbering may be stale elsewhere. 0.9 (introspection + console + D1–D3), 0.10 (unified start / dynamic supervisor / OTP supervisor alignment), and 0.11 (admission) replace the old 0.9/0.10 split. Grep ROADMAP.md and docs/ for the previous mapping so ranks 12a, 13–14, and 17–19 are not still tied to 0.9 or 0.10.

Changelog bullet is hard to parse. Split the three facts (smaller releases, per-feature implementation bar, coverage scope) into separate sentences.

The new “Implementing a release” bar (design test plan, trace-tied model, planted bugs, no-cost bench, independent review, soak/canary) is clear and consistent with the stated reason for splitting 0.8. “spawned-ctl attach on every binary that runs an ActorSystem” is a strong product commitment; worth an explicit opt-out in the design if some embeddings must not open a control socket.

---

Automated review by Grok (xAI) · grok-4.7 · custom prompt

[github-actions]

🤖 Claude Code Review

Review of PR #269 (ROADMAP and CHANGELOG only)

This PR changes only documentation, so there are no correctness or concurrency concerns. Splitting the old 0.9 into two single-theme releases is a sensible change. My comments are about consistency and precision.

ROADMAP.md

1. Stale version references elsewhere (lines ~95-97)

• The admission package moved from 0.10 to 0.11. Unified start and the dynamic supervisor moved from 0.9 to 0.10.
• Grep ROADMAP.md, docs/design/* and docs/TESTING.md for remaining 0.9 and 0.10 references. Likely places are the per-rank target column for ranks 12a, 13, 14, 17, 18, 18a and 19, and any "After v0.9" gate rows.
• As written, the release list may now contradict other tables in the same file.

2. Link targets (lines 95-96)

• Confirm that docs/design/introspection.md and docs/design/unified-start.md exist on main. They are not added in this diff, and relative links to missing files render as broken.
• The dynamic supervisor design links to PR design: dynamic supervisor (rank 12a) #245 instead of a design doc. If design: dynamic supervisor (rank 12a) #245 is still open, calling it "approved" is premature. Once it merges, link a docs/design/dynamic-supervisor.md instead, for consistency with the other two designs.

3. spawned-ctl attach on every binary (line 95)

• "on every binary that runs an ActorSystem" reads as an attach endpoint that is always on.
• An introspection console that can call get_state is a security and data-exposure surface. Say explicitly whether it is opt-in, for example via a feature flag or a builder option, and how it is bound (a Unix socket with file permissions, or something else).
• If the design doc already settles this, mirror it here in a few words.

4. Undefined identifiers (lines 95, 105)

• "D1-D3": link to the audit section that defines these IDs. The old wording did not need IDs.
• "model-sync TLC-only allowlist": add a link to TESTING.md or wherever the allowlist is defined. Readers outside the team will not know what this means.
• "rank 8" for the stall watchdog: check that this matches the Priorities table.

5. Wording (lines 101-110)

• "each review round found new findings" is redundant. Suggest: "each review round surfaced new issues where they interacted."
• "a no-cost benchmark against the previous release" is ambiguous. Suggest: "a benchmark showing no regression against the previous release."
• "the next release's work proceeds in parallel on branches" sits a little awkwardly next to "one theme each". Consider stating that only one release is in review at a time, if that is the intent.

6. Formatting

• 0.9 and 0.10 now have bold titles (**0.9: introspection and the attachable console**), but 0.8 and 0.11 do not. Either give every entry a theme title or none.
• The new paragraphs are hard-wrapped at about 78 columns, while the surrounding bullets and table rows use one long line each. Matching the existing style keeps future diffs clean.

CHANGELOG.md

7. Coverage wording mismatch (line 8)

• The changelog says "widened to the whole crate over the simulated suites".
• The ROADMAP row says "measured over the ordinary and the simulated (--cfg spawned_sim) suites".
• Make the two agree, for example: "over both the ordinary and simulated suites".
• Also, "the whole crate" is ambiguous in a workspace. Name the crate (spawned-concurrency?) or say "the workspace".

8. Unclear whether the coverage change is real or planned

• If this PR only updates the plan (the row is still "Ongoing"), the changelog should say the roadmap now plans the wider report.
• As written, it could be read as saying the CI coverage job was actually changed.

Summary

The restructuring is good. Before merging:

• Fix any stale version references elsewhere in the docs (item 1).
• Verify the link targets (item 2).
• Clarify that the attach console is opt-in (item 3).

The remaining items are wording and formatting polish.

---

Automated review by Claude · claude-opus-5-5 · custom prompt

[github-actions]

🤖 Codex Code Review

• [P2] Broken design link — ROADMAP.md:96: docs/design/unified-start.md does not exist. Add the design document or link to its actual location so readers can access the approved design.
• [P3] Stale release reference — ROADMAP.md:118: “the 0.10 admission package” should now say “0.11,” matching the revised release schedule.

This PR only changes documentation; no runtime, concurrency, or memory-safety concerns.

---

Automated review by OpenAI Codex · gpt-6.1-sol · custom prompt

[github-actions]

🤖 Kimi Code Review

No blocking issues. This is a documentation-only change, so there are no runtime, concurrency, or memory-safety concerns.

Minor wording issue:

• CHANGELOG.md:5-8: “coverage report widened to the whole crate over the simulated suites” is awkward and less precise than the ROADMAP text. Consider: “the code coverage report widened to the whole crate across both ordinary and simulated (--cfg spawned_sim) test suites.” This also avoids implying that only simulated suites are measured.
• ROADMAP.md:108: “a no-cost benchmark” is ambiguous. If the intent is to guard against regressions, “a benchmark comparison against the previous release showing no performance regression” would be clearer.

Otherwise, the release reordering and coverage description are internally consistent.

---

Automated review by Kimi (Moonshot AI) · kimi-k3.1-pro · custom prompt