docs: remove stale handoffs and consolidate current guidance

[kalepail]

Summary

• Replace the completed phase plan with current scope and links to each authoritative document.
• Remove stale deployment summaries and completed decisions from the handoff.
• Keep unresolved owner decisions, paid-evaluation limits, and source acceptance requirements.
• Correct stale counts, historical status claims, and the inventory cache description.
• Keep runtime code, generated artifacts, source pins, golden answers, and finding records unchanged.

Coverage and preservation

The audit inventoried all 1,185 tracked Markdown and text documents.
It used 51 current guidance documents as cleanup candidates.
It preserved 1,134 evidence, generated, contract, and historical documents through explicit exclusions.
This was not a line-by-line truth review of the historical corpus.

The cleanup changes 13 documents and removes about 500 net lines.
No complete document was deleted.
Git history retains the removed duplicate guidance.

The relative Markdown file-link scan found no broken links in the current guidance group.
Three historical references remain unchanged: two quoted upstream anchors and one archived corpus reference.
The scan does not validate external URLs or Markdown heading anchors.

Verification

• npm run typecheck: passed.
• npm test: 2,074 tests passed across 114 files.
• npm run build: passed.
• git diff --check: passed.
• Secret scans: passed for the tracked tree and staged changes.
• Independent preservation review: Grok 4.6 high; the budget restriction is restored.
• Independent standards review: Sol high; the signature description and line-count claim are corrected.
• GitHub CI and CodeQL: all four checks passed at ac1769f.

Both reviewers differ from the author and orchestrator.
The reviews checked preservation and documentation contracts, not product design.
The coverage review withdrew its assumption that unchanged files were unread files.

No paid evaluation, merge, or deployment occurred for this documentation change.

[Copilot]

🟢 Approval recommended

Only a non-blocking documentation nit remains; no approval-blocking issues were identified.

Pull request overview

This documentation-only PR replaces stale planning and handoff history with current guidance while preserving runtime code, evidence, and active decisions.

Changes:

• Updates scope, ownership, architecture, and authoritative links.
• Corrects inventory and evaluation documentation.
• Clarifies historical research and current priorities.

File summaries

File	Summary
PLAN.md	Current scope and maintenance guidance
inventory/README.md	Corrected catalog cache behavior
ideas/README.md	Updated research-note index
ideas/per-user-mcp-observability.md	Clarified historical status
ideas/observability-r2-retention.md	Clarified historical retention research
ideas/architecture-explorations.md	Consolidated experiment status
eval/README.md	Updated evaluation guidance
eval/qa/README.md	Clarified corpus lifecycle guidance
eval/playground/README.md	Removed stale battery count
eval/discovery/README.md	Clarified seed-pool provenance
ARCHITECTURE.md	Updated current-state framing
.agents/TODO.md	Simplified maintenance references
.agents/NEXT.md	Updated current priorities and decisions

Review details

Suppressed comments (1)

PLAN.md:22

• This ownership sentence contradicts the map in §6: src/server.ts/src/auth/ own MCP authentication, while manifest argument validation is in the policy/executor layers and src/adapters/ own upstream traffic. Calling all of these responsibilities “Host adapters” can send maintainers to the wrong owner; split the host-side layers from service adapters here.

Host adapters own service traffic, authentication, argument validation, and secrets.

• Files reviewed: 13/13 changed files
• Comments generated: 0
• Review effort level: Lite

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.