ietf-draft-analyzer/data/reports/platform-improvement-plan.md

Platform Improvement Plan: IETF Draft Analyzer → Standards Intelligence Platform

Generated 2026-03-07 — Based on full codebase audit and architectural analysis

---

Current State Summary

Dimension	What Exists	Assessment
Data	434 drafts, 557 authors, 419 ideas, 11 gaps, 4231 refs to 694 RFCs	Strong foundation
CLI	20+ commands (fetch, analyze, embed, ideas, gaps, report, viz, wg, etc.)	Feature-rich
Web UI	16 Flask pages with D3.js/Plotly visualizations	Good but disconnected
Pipeline	Observatory class with multi-source support	Built but manual
Multi-SDO	IETF complete, W3C fetcher written but unused	Partially built
Tests	Zero tests	Critical gap
Deployment	Manual Python setup	No containerization

The Transformation

From: A powerful CLI analysis tool that an expert runs manually To: A living intelligence platform that monitors, alerts, and answers questions

---

Phase 1: Quick Wins (All Parallelizable)

1.1 Global Search [S: 3-5h]

Add a unified search bar to the web UI that queries across drafts (FTS5), ideas, authors, and gaps simultaneously. Currently search is isolated to the drafts page.

• Files: src/webui/app.py (new /search route), src/webui/data.py (new global_search()), src/webui/templates/base.html (search in sidebar), new search_results.html
• Why first: Most common user action on any data platform

1.2 REST API [S: 2-3h]

Expose all existing data.py functions as JSON endpoints. Only 3 API endpoints exist today — extend to all 15+ data views.

• Files: src/webui/app.py (add /api/ideas, /api/gaps, /api/ratings, /api/timeline, /api/landscape, /api/similarity, /api/drafts/<name>, etc.)
• Why: Enables programmatic access, third-party tools, decouples data from presentation

1.3 Export (CSV/JSON) [S: 3-4h]

Add export buttons to web UI pages + ietf export CLI command.

• Files: src/ietf_analyzer/cli.py (new export command), API endpoints get ?format=csv support
• Depends on: 1.2

1.4 Annotation System [M: 4-6h]

Add private notes and custom tags per draft, persisted in DB.

• Files: src/ietf_analyzer/db.py (new annotations table), src/webui/app.py (POST endpoint), src/webui/templates/draft_detail.html (inline edit), src/ietf_analyzer/cli.py (new annotate command)
• Why: Analysts need to layer their own context onto the data

1.5 Test Suite Foundation [M: 6-8h]

Create pytest infrastructure with ~30 tests covering DB layer, models, and web data functions.

• Files: new tests/conftest.py, tests/test_db.py, tests/test_models.py, tests/test_data.py, update pyproject.toml
• Why: Zero tests today. Every future change benefits from a safety net

---

Phase 2: Core Platform

2.1 Semantic Search / "Ask" [M: 1-2 days] — THE SIGNATURE FEATURE

Natural language queries: "Which drafts address agent authentication?" → synthesized answer with citations.

• Embed the query via Ollama, compute cosine similarity against all 434 draft embeddings
• Merge with FTS5 keyword results using reciprocal rank fusion
• Optionally synthesize answer via Claude with top-K context
• Files: new src/ietf_analyzer/search.py, src/ietf_analyzer/cli.py (ietf ask), src/webui/app.py (/ask route), new template

2.2 Competitive Landscape Mapping [M: 1-2 days]

Auto-detect and compare competing proposals in the same problem space.

• Group drafts by similarity clusters, enrich with rating comparisons and WG adoption status
• Show head-to-head comparisons: where they agree, where they diverge
• Files: new src/ietf_analyzer/competition.py, src/webui/app.py (/competition route), new template

2.3 Standards Readiness Scoring [M: 1 day]

Composite 0-100 "readiness" score: WG adoption, revision count, reference density, cited-by count, author track record.

• Files: new src/ietf_analyzer/readiness.py, update src/webui/templates/draft_detail.html (gauge chart), update drafts listing

2.4 Scheduled Updates + Pipeline Health [M: 1 day]

Cron-based auto-fetch using existing Observatory, plus monitoring dashboard.

• Files: new scripts/scheduled-update.sh, enhance src/webui/templates/monitor.html (stage breakdown, cost tracking, failure log)

2.5 Comparison View [M: 1 day]

Side-by-side comparison of 2+ drafts with rating radar overlay, shared/unique ideas, shared/unique references.

• Files: src/webui/app.py (/compare?drafts=...), src/webui/data.py (get_comparison_data()), new template, add checkboxes to drafts listing

---

Phase 3: Intelligence

3.1 Trend Forecasting [L: 2-3 days]

Predict which areas will grow based on submission velocity, revision activity, WG adoption signals.

• Per-category momentum signals, linear/exponential extrapolation, "Hot/Cooling/Emerging" tags
• Files: new src/ietf_analyzer/forecasting.py, new web page /trends

3.2 Change Detection & Diffing [L: 2-3 days]

Track what changed between draft revisions, summarize changes via Claude.

• New draft_revisions table to archive old versions
• Section-level diff with Claude-generated change summaries
• Files: new src/ietf_analyzer/diff.py, update src/ietf_analyzer/db.py, src/ietf_analyzer/fetcher.py
• Depends on: 2.4 (scheduled updates to catch new revisions)

3.3 Citation Graph [M: 1-2 days]

Visual dependency tree from the 4231 existing cross-references to 694 RFCs.

• D3.js force-directed graph (pattern from authors.html), PageRank-style influence scores
• Files: src/webui/data.py (get_citation_graph()), new template citations.html

3.4 Newsletter Generation [M: 1-2 days]

Automated weekly/monthly digest with new drafts, significant changes, trend shifts.

• Files: new src/ietf_analyzer/newsletter.py, src/ietf_analyzer/cli.py (ietf newsletter)
• Depends on: 3.2 (for change content)

---

Phase 4: Scale

4.1 Docker Deployment [S: 3-4h]

Dockerfile + docker-compose.yml with Flask app + Ollama.

4.2 Complete W3C Integration [M: 1 day]

Wire existing W3C fetcher into Observatory pipeline, make prompts source-aware, add source filter to web UI.

4.3 IEEE + 3GPP Sources [L: 3-5 days]

New source fetchers following SourceFetcher protocol. Depends on 4.2 validating the pipeline.

4.4 Cross-SDO Analysis [L: 2-3 days]

Compare work across standards bodies. Embedding similarity between IETF/W3C/IEEE specs, gap analysis for topics only one body covers.

4.5 Plugin Architecture [L: 2-3 days]

Formalize extension points for sources, analyzers, report types via Python entry points.

---

Recommended Agent Team Assignments

Immediate Sprint (Phase 1 — all in parallel)

Agent	Items	Focus
Coder A	1.1 + 1.2	Global search + REST API
Coder B	1.4 + 1.3	Annotations + Export
Coder C	1.5	Test suite foundation

Next Sprint (Phase 2 — mostly parallel)

Agent	Items	Focus
Coder A	2.1	Semantic search / Ask (signature feature)
Coder B	2.2 + 2.5	Competition mapping + Comparison view
Coder C	2.3 + 2.4	Readiness scoring + Pipeline health

---

Impact vs Effort Matrix

HIGH IMPACT
    |
    |  2.1 Ask          3.2 Diffing
    |  1.1 Search       3.1 Forecasting
    |  2.2 Competition  4.4 Cross-SDO
    |  2.5 Compare
    |  1.2 API
    |  2.3 Readiness    3.3 Citations
    |  1.4 Annotations  3.4 Newsletter
    |  1.3 Export       4.2 W3C
    |  1.5 Tests        4.1 Docker
    |                   4.5 Plugins
    |                   4.3 IEEE/3GPP
LOW IMPACT
    +------------------------------------------
    LOW EFFORT                      HIGH EFFORT