docs: align roadmap with focused stories and evidence limits

[mohanagy]

Summary

The roadmap presented superseded recovery work and completed research as the next execution queue. Update docs/roadmap.md to the focused stories, the initial TypeScript/Node workflow and the limits of the available evidence.

The page separates source correctness, context usefulness and completed-task benefit. It records #753's completed inventory and pinned control/candidate roles, keeps #739 blocked and #710's remaining session failures visible, and leaves #754/#756 deferred. The previous contributor roadmap remains unchanged in its historical archive; the September 5 decision has an immutable history link.

Acceptance criteria

• Describe the initial workflow, with ordinary follow-up source reading allowed and counted.
• Distinguish the three evidence claims and preserve failed/incomplete outcomes without trust percentages or superiority claims.
• Link the focused owners and their current dispositions.
• Archive superseded planning without presenting roadmap: separate maintenance, evidence use, and Madar product qualification #740 or completed research: define an evidence-use experiment and separate acceptance metrics #741/research: build the offline evidence-reconciliation evaluator prototype #743 as the next execution program.

Validation

• Read-only review confirms all four documentation criteria. Head 80c8910c60bc49ac80366d475be2a9fe068121f4 has the same tree as the previously reviewed 0f879b8d: 2e41f4d3615a892303daccfa6b91da938b18a1e7.
• Historical archive unchanged; the diff against next changes only docs/roadmap.md. git diff --check passed.
• Manual CI run 34516683472, attempt 2 passed all six jobs. Its first macOS Node 20 attempt failed with a child-process EPIPE; the unchanged-head retry passed. The original failure remains recorded and its cause is not declared fixed.
• The manual run did not clear the skipped PR-workflow gate. An empty commit without the skip instruction started normal PR CI run 34519554504. All six required PR checks passed on the first attempt at 80c8910c60bc49ac80366d475be2a9fe068121f4.

Merged into next as d1723f4b7bc6af989073bef899e953f8554aa340. The merge changes only docs/roadmap.md, whose blob matches the reviewed PR. Post-merge CI run 34521062988 passed all six jobs on the merged commit on the first attempt. Source implementation, new model experiments and releases remain outside this change.

Summary by CodeRabbit

• Documentation
  • Updated the project roadmap with a dated, evidence-based work queue.
  • Clarified supported TypeScript and Node.js workflows.
  • Organized findings by source correctness, context usefulness, and completed-task benefit.
  • Documented unresolved evaluation findings, scoped issues, current dispositions, simplification boundaries, comparison criteria, ownership limits, and completed work.
  • Preserved the historical roadmap archive and corrected its closing formatting.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Team

Run ID: 0bf6cb36-9938-4402-89ac-222369270cb8

📥 Commits

Reviewing files that changed from the base of the PR and between 2b14450 and 0f879b8.

📒 Files selected for processing (1)

• docs/roadmap.md

Included review availability: 0 reviews are currently available. Your included PR review attempts over the past 7 days set your current allowance at 1 review per hour.

---

📝 Walkthrough

Walkthrough

The roadmap now defines the supported workflow, evidence limits, issue dispositions, comparison criteria, execution boundaries, completed work, and preserved historical archive.

Changes

Roadmap documentation

Layer / File(s)	Summary
Current roadmap and archive structure docs/roadmap.md	The roadmap adds a dated work queue with scope, evidence limits, issue dispositions, comparison criteria, execution boundaries, and completed work. It also closes the preserved historical archive with </details>.

Estimated code review effort: 1 (Trivial) | ~3 minutes

Merge Risk: ⚪ Minimal · up to 80c89

This documentation-only update clarifies roadmap scope and preserves historical planning without changing product behavior or runtime operations.

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check. Docstring coverage is scoped to functions touched by this diff. Analyzed 0 functions across 0…
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Title check	✅ Passed	The title is concise, specific, and accurately describes the main documentation change: aligning the roadmap with focused stories and evidence limits.
Description check	✅ Passed	The description provides a clear summary, acceptance criteria, validation details, testing status, scope boundaries, and related issue references. It does not use the exact Testing, Checklist, and Rel…

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch docs/740-roadmap-refinement

---

_{Comment @coderabbitai help to get the list of available commands.}

[mohanagy]

Required CI at 0f879b8d635950673530d5550124ac7e2bcf22cb recorded a macOS Node 20 failure in attempt 1, job 103003721929.

tests/unit/host-discovery-all-surfaces.test.ts failed its Gemini current-v2/tombstone case with spawnSync ... node EPIPE. The test helper supplies JSON stdin and throws result.error at lines 131–136, before the guidance/classification assertions. The generated Gemini program reads its payloads from argv and does not consume that stdin. This is consistent with a child-process I/O/lifecycle failure; the cause is not established.

The job reported 328 passing files and one failing file; 5,681 passing tests, one failure, one expected failure and 20 skips. These are the original attempt's results and remain part of the record.

The PR changes only docs/roadmap.md; the affected test/helper and generated-hook source are unchanged from next. One same-head retry of this failed job is planned after the original matrix finishes. A retry pass would not mean the cause was fixed or prove an infrastructure-only explanation. No source/test change or broader repair is included.