Real Archive Attention Queue

Date: 2026-06-28

What Changed

The dashboard was redesigned again after Arijit rejected the prototype-like weekly brief. The new direction uses actual archived CI reports only. No mock findings and no fake buttons are allowed.

Actual Data Used

The local archive contains:

• Daily briefs:
• 2026-06-17.md
• 2026-06-20.md
• 2026-06-27.md
• Weekly syntheses:
• 2026-06-17-weekly.md
• 2026-06-18-weekly.md
• 2026-06-27-weekly.md

This is a sparse archive across 2026-06-17 to 2026-06-27, not fourteen daily reports.

Dashboard IA

The homepage is now an attention queue:

1. The strongest actual finding: AWS OpenSearch Serverless / agent skills pressure on Algolia's agentic search story.
2. The second finding: Constructor and Bloomreach customer proof needs sales validation.
3. The repeated pattern: AI search packaging across MCP, agents, commerce reasoning, personalization, and product discovery.

The right rail explains reliability:

• Uses real archived briefs.
• Archive is sparse.
• Public-source only.
• Local ci.sqlite ledger is missing, so this view is generated from report files.

Files Updated

• Standalone repo:
• apps/dashboard/public/index.html
• apps/dashboard/scripts/verify-static.js
• apps/dashboard/public/archive/*.md
• Local vault dashboard artifact:
• artifacts/competitive-research/dashboard/index.html

Validation

• npm run dashboard:build passed.
• npm run test passed: 18 tests.
• Archive files were copied and verified.
• Headless Chrome desktop and mobile screenshots rendered.

Next

The next build should stop hand-authoring the dashboard and generate the attention queue from report files or the ledger:

• Parse archived Markdown reports into structured findings.
• Rank findings by strategic importance, not raw signal score.
• Preserve the source report and source URL for every finding.
• When ci.sqlite is restored locally and on VPS, drive the dashboard from ledger records plus report archive.