CODITECT Research Artifact Organization — Taxonomy & Information Architecture Proposal

Status: Proposed — 2026-02-16 Author: Claude (Sonnet 4.5) Audience: CODITECT Contributors, Documentation Librarians, Framework Architects

Executive Summary

CODITECT has accumulated 98 research directories (6.1GB, 1229+ markdown files, 384 PDFs) in a gitignored staging area (analyze-new-artifacts/). Only a single "assessment document" gets promoted to permanent storage (internal/analysis/), leaving SDDs, TDDs, ADRs, C4 diagrams, JSX dashboards, and source materials orphaned.

This proposal establishes:

6-category research taxonomy with clear inclusion criteria
4-stage artifact lifecycle (Staging → Analysis → Integration → Archive)
Promotion criteria matrix defining when/how artifacts move between stages
Permanent directory structure replacing ad-hoc organization
Manifest system tracking research lineage and outputs
Integration with ADR-206 pipeline (autonomous research workflow)

Impact:

87% reduction in staging clutter (98 dirs → 13 permanent topics)
Zero loss of valuable artifacts — systematic promotion replaces deletion
Searchable knowledge base — all research findable via manifests and indexes
Pipeline integration — ADR-206 outputs directly to permanent locations

1. Problem Statement

1.1 Current State Analysis

Staging Area (analyze-new-artifacts/):

98 directories containing:
- 1,229 markdown files
- 384 PDF papers
- 70 JSX dashboards
- ~200+ ADRs (duplicates, drafts, production)
- 6.1GB total size
Naming inconsistency: coditect-, CODITECT-, mixed case, duplicates (*-DUPLICATE)
No lifecycle management: Files stay forever or get manually deleted
No manifest tracking: Unknown provenance, no lineage

Permanent Storage:

internal/analysis/ (20 directories) — assessment documents only
internal/research/ (7 directories) — strategic research
Gap: No permanent home for SDDs, TDDs, ADRs, dashboards, source materials

Workflow Breakdown:

Research Session → Staging Area (gitignored) → ??? → Lost Forever
                         ↓
                   Single Assessment Doc → internal/analysis/

What Gets Lost:

Software Design Documents (SDDs)
Technical Design Documents (TDDs)
Architecture Decision Records (ADRs)
C4 architecture diagrams
JSX dashboards (React components)
Glossaries and reference materials
Mermaid diagrams
Source materials (PDFs, transcripts, blog posts)
Follow-up prompts and ideation

1.2 ADR-206 Context

The ADR-206 Autonomous Research Pipeline (approved 2026-02-16) defines a 4-phase multi-agent workflow producing:

Phase 1 Outputs (9 artifacts):

1-2-3-detailed-quick-start.md
coditect-impact.md
executive-summary.md
sdd.md
tdd.md
c4-architecture.md
adrs/ (3-7 ADRs)
glossary.md
mermaid-diagrams.md

Phase 2 Outputs (4-6 dashboards):

tech-architecture-analyzer.jsx
strategic-fit-dashboard.jsx
coditect-integration-playbook.jsx
executive-decision-brief.jsx
competitive-comparison.jsx (extended)
implementation-planner.jsx (extended)

Phase 3 Outputs:

follow-up-prompts.md (15-25 prompts across 6 categories)

Phase 4:

Handoff to /new-project for project genesis

Problem: ADR-206 defines artifact CREATION but NOT artifact PROMOTION to permanent storage. Without this proposal, ADR-206 outputs would remain in analyze-new-artifacts/ indefinitely.

2. Research Taxonomy (6 Categories)

2.1 Taxonomy Design Principles

Purpose-driven — Categories reflect research intent, not format
Mutually exclusive — Each research topic belongs to exactly one category
Lifecycle-aware — Categories have distinct promotion criteria
Scalable — Can accommodate 100+ research topics without restructuring

2.2 Category Definitions

Category 1: Technology Evaluation

Purpose: Assess external tools, frameworks, APIs, or platforms for CODITECT integration.

Characteristics:

Focuses on specific technology (e.g., Agent Labs, CopilotKit, Codex)
Produces integration feasibility assessment
Results in Go/No-Go decision (often captured as ADR)
Time-bound (evaluation period: 1-7 days)

Example Topics:

agent-labs, copilotkit, motia, openclaw, codex
opencode, agent-zero, unified-studio, bookmarks-buku
moltbot, clawdrop, paperbanana, plugins

Typical Artifacts:

SDD (integration design)
TDD (API specifications)
ADRs (adoption decision)
Executive summary
CODITECT impact analysis
JSX dashboards (competitive comparison, tech architecture analyzer)

Promotion Criteria (see Section 3):

Executive summary → internal/analysis/{topic}/
ADRs → internal/architecture/adrs/ (if adopted) or internal/architecture/adrs/rejected/
SDD/TDD → internal/architecture/integrations/{topic}/ (if adopted)

Category 2: Academic Research

Purpose: Extract patterns, algorithms, or methodologies from academic papers for CODITECT implementation.

Characteristics:

Source: arXiv, academic journals, conference proceedings
Focuses on research findings, not commercial products
Often multi-paper synthesis
Results in implementation roadmap or capability enhancement

Example Topics:

scaling-agent-systems (arXiv 2512.08296v2)
recursive-llms, continual-learning, context-graph
parallel-agent-reinforcement-learning
ambiguity-and-intent, ai-guardrails
agentic-paradigms-healthcare, prompt-repetition
consequence-aware-autonomous-execution

Typical Artifacts:

Paper annotations (PDF + markdown conversion via UDOM)
Executive summary (paper synthesis)
Implementation recommendations
ADRs (if pattern adopted)
Glossary (domain terminology)
Mermaid diagrams (visual synthesis)

Promotion Criteria:

Executive summary → internal/analysis/academic/{topic}/
Paper PDFs + markdown → internal/research/academic/{topic}/papers/
ADRs → internal/architecture/adrs/ (if pattern implemented)
Implementation recommendations → internal/architecture/academic-patterns/

Category 3: Competitive Intelligence

Purpose: Monitor competitors, market positioning, and differentiation strategies.

Characteristics:

Focus: competitor products, pricing, features, market share
Sources: competitor websites, demos, press releases, investor updates
Time-sensitive (market changes frequently)
Results in positioning strategy or feature prioritization

Example Topics:

anthropic-cowork-impact, palantir, kimi-2.5
microsoft-fabric, ibm-ai, eric-schmidt-ai
dylan-davis (thought leader analysis)

Typical Artifacts:

Competitive comparison matrix
Feature gap analysis
Pricing analysis
Market positioning recommendations
Executive brief (board-level)
JSX dashboard (competitive-comparison.jsx)

Promotion Criteria:

Executive summary → internal/research/market-research/competitors/{company}/
Competitive matrix → internal/research/market-research/competitive-analysis/
Dashboards → internal/research/market-research/dashboards/
Note: Competitive intel ages quickly — auto-archive after 12 months unless refreshed

Category 4: Business & Market Research

Purpose: Validate market opportunity, business models, pricing, and go-to-market strategies.

Characteristics:

Focus: TAM/SAM/SOM, pricing models, customer economics, ROI
Sources: market research reports, case studies, financial models
Results in business case or GTM strategy
Board-level decision support

Example Topics:

product-market-fit, tiny-seed, runway-mentor-deck
financial-model, value-proposition, use-cases
hi-tech-ventures, roblox-market (market sizing analogs)

Typical Artifacts:

Business case documents
Financial models (Excel/JSON)
Executive summaries (1-page)
Market sizing analysis
Pricing strategy
Customer success economics
JSX dashboards (strategic-fit-dashboard.jsx, executive-decision-brief.jsx)

Promotion Criteria:

Business cases → internal/research/business/cases/{topic}/
Financial models → internal/research/business/models/{topic}/
Market sizing → internal/research/market-research/sizing/
Executive summaries → internal/research/business/executive-briefs/

Category 5: Domain Research

Purpose: Deep-dive into specific industries (healthcare, biotech, legal, security) for domain-specific feature development.

Characteristics:

Focus: industry workflows, compliance, regulations, terminology
Results in domain-specific product features or compliance requirements
Often drives Track G (DMS Product) or Track O-AA (PCF Business) work

Example Topics:

bioscience-workorders, LIMS-sample-receiving, healthcare
regulatory-frameworks, ai-risk-management, legal-contracts
license-management, c3pao-accreditation, zero-trust-security

Typical Artifacts:

Domain glossary (terminology)
Workflow diagrams (Mermaid)
Compliance requirements (checklists)
SDD (domain-specific features)
TDD (data models for domain)
ADRs (compliance decisions)

Promotion Criteria:

Glossaries → internal/research/domain/{industry}/glossary.md
Workflow diagrams → internal/research/domain/{industry}/workflows/
Compliance requirements → internal/research/domain/{industry}/compliance/
SDD/TDD → internal/architecture/domain/{industry}/
ADRs → internal/architecture/adrs/ (tagged with domain)

Category 6: Process & Internal Research

Purpose: Improve CODITECT's own processes, infrastructure, or framework capabilities.

Characteristics:

Focus: internal tools, developer experience, automation
Results in framework enhancements or process improvements
Often drives Track H (Framework) or Track C (DevOps) work

Example Topics:

system-prompt, master-prompt, conflict-avoidance
decision-rights, new-project-initiation, method-for-analysis
core-process-framework, docker-registry, docusaurus-search
new-installation-errors, ai-applicability-occupations

Typical Artifacts:

Process documentation
System prompt updates
Framework enhancement proposals
Infrastructure analysis
Developer guides
ADRs (framework decisions)

Promotion Criteria:

Process docs → internal/process/{topic}/
System prompts → .coditect/prompts/{topic}/ (if adopted)
Framework proposals → internal/architecture/framework-enhancements/
Infrastructure analysis → internal/research/infrastructure/
ADRs → internal/architecture/adrs/

2.3 Special Categories (Non-Research Content)

Media & Transcripts

Nature: Raw source materials, not research outputs.

Content: YouTube transcripts, podcast transcripts, conference recordings.

Examples:

davos-economic-forum, AI-HOUSE-DAVOS, a16z-youtube
lex-clips, all-in-podcast, nate-b-jones-youtube
internal-meeting-notes

Lifecycle:

Stay in staging during active research
Move to internal/research/source-materials/transcripts/{topic}/ if referenced by permanent analysis
Otherwise delete after research completion (regenerable via YouTube API)

UI/UX & Visual

Nature: Screenshots, images, design mockups.

Content: Browser screenshots, profile images, UI analysis.

Examples:

screenshots, browser-analysis, browser-screenshots
analyze-ui-images, profile-images, ui-ux-agent-design

Lifecycle:

Visual references for design work
Move to internal/research/design/{topic}/ if part of design system research
Otherwise delete (transient reference materials)

3. Artifact Lifecycle (4 Stages)

3.1 Lifecycle Model

Stage 1: STAGING
  ↓
Stage 2: ANALYSIS
  ↓
Stage 3: INTEGRATION
  ↓
Stage 4: ARCHIVE

3.2 Stage 1: STAGING

Location: analyze-new-artifacts/{topic}/

Status: Gitignored (ephemeral, not tracked)

Purpose: Active research workspace for artifact generation.

Duration: 1-30 days (typical: 1-7 days)

Contents:

Source materials (PDFs, transcripts, web crawls)
Draft artifacts (SDDs, TDDs, ADRs)
Generated outputs (dashboards, diagrams)
Work-in-progress notes

Activities:

Research pipeline execution (ADR-206)
Manual research sessions
Multi-agent artifact generation
Quality validation iterations

Exit Criteria:

Research complete (all planned artifacts generated)
Quality gate passed (validation score ≥0.7)
Promotion decision made (see Section 3.4)

Automation:

ADR-206 pipeline auto-creates staging directories
Manifest auto-generated on pipeline completion
Quality validation run before promotion

3.3 Stage 2: ANALYSIS

Location: internal/analysis/{category}/{topic}/

Status: Git-tracked (permanent record)

Purpose: Refined assessment documents for future reference.

Duration: Indefinite (permanent unless superseded)

Contents (Promoted from Staging):

Executive summary (primary artifact)
Assessment document (integration feasibility, impact analysis)
Key findings summary
Recommendations
References to source materials
Manifest (tracking research lineage)

NOT Included (See Stage 3 for these):

SDDs/TDDs (go to internal/architecture/)
ADRs (go to internal/architecture/adrs/)
Dashboards (go to stage-specific locations)
Source materials (stay in staging or move to source-materials/)

Promotion Criteria from Staging:

✅ Research concluded with actionable findings
✅ Assessment document written (not just pipeline outputs)
✅ Quality validated (no placeholder content)
✅ References verified (links to sources work)
✅ Manifest created (see Section 5)

Example Structure:

internal/analysis/technology-evaluation/agent-labs/
├── agent-labs-scaling-assessment-2026-02-16.md  # Primary assessment
├── executive-summary.md                          # Promoted from pipeline
├── recommendations.md                            # Actionable next steps
├── MANIFEST.md                                   # Research lineage
└── README.md                                     # Navigation document

3.4 Stage 3: INTEGRATION

Location: Varies by artifact type (see Section 4)

Status: Git-tracked (production architecture/docs)

Purpose: Artifacts integrated into CODITECT's production architecture or documentation.

Duration: Indefinite (permanent, versioned)

Contents (Promoted from Analysis):

SDDs → internal/architecture/integrations/{topic}/ or domain/{industry}/
TDDs → internal/architecture/integrations/{topic}/ or domain/{industry}/
ADRs → internal/architecture/adrs/ (numbered, indexed)
C4 diagrams → internal/architecture/c4-diagrams/integrations/{topic}/
Business cases → internal/research/business/cases/{topic}/
Financial models → internal/research/business/models/{topic}/
Domain glossaries → internal/research/domain/{industry}/glossary.md
Dashboards → (depends on adoption — see Section 4.6)

Promotion Criteria from Analysis:

✅ Decision made (Go decision in ADR or executive summary)
✅ Artifact ready for production reference (no draft content)
✅ Properly formatted (follows CODITECT standards)
✅ Numbered/indexed appropriately (ADRs, SDDs)
✅ Cross-referenced (links from analysis docs)

Integration Tracking:

Analysis document updated with "Integration Status" section
Manifest updated with integration locations
ADR recorded for any architecture decisions

3.5 Stage 4: ARCHIVE

Location: internal/research/archive/{category}/{topic}/

Status: Git-tracked (historical record)

Purpose: Historical research that is no longer relevant but should be preserved for reference.

Duration: Permanent (never deleted)

Archive Triggers:

Superseded by newer research (e.g., competitive intel updated)
Technology deprecated (e.g., evaluated tool no longer exists)
Decision reversed (e.g., Go decision later changed to No-Go)
Time-sensitive research aged out (e.g., 12-month-old market data)

Archive Process:

Move entire analysis directory to archive location
Add status: archived to frontmatter
Add superseded_by: or archive_reason: field
Update manifest with archive date and reason
Create redirect in original location (README pointing to archive)

Example:

internal/research/archive/competitive-intelligence/palantir-2025/
├── palantir-competitive-analysis-2025-06-15.md  # Original assessment
├── MANIFEST.md                                   # Preserved lineage
└── ARCHIVE-NOTICE.md                             # Why archived, when, by whom

3.6 Lifecycle Decision Matrix

Artifact Type	Staging → Analysis	Analysis → Integration	Analysis → Archive
Executive Summary	Always	Never (stays in analysis)	If superseded
Assessment Doc	Always	Never (stays in analysis)	If superseded
SDD	Sometimes*	If Go decision	N/A
TDD	Sometimes*	If Go decision	N/A
ADR	Sometimes*	If decision finalized	If decision reversed
C4 Diagrams	Sometimes*	If Go decision	N/A
Business Case	Always	Always	If superseded
Financial Model	Always	Always	If superseded
Glossary	Sometimes*	If domain research	N/A
Dashboards	Rarely**	Rarely**	N/A
Source Materials	Rarely***	Never	N/A

*Promoted to analysis only if referenced by assessment document. **Dashboards typically stay in staging unless adopted for production use. ***Source materials referenced by URL/path, not copied to analysis.

4. Permanent Directory Structure

4.1 Top-Level Organization

internal/
├── analysis/                    # Stage 2: Refined assessments
│   ├── technology-evaluation/
│   ├── academic/
│   ├── competitive-intelligence/
│   ├── business-market/
│   ├── domain/
│   └── process-internal/
│
├── architecture/                # Stage 3: Integrated architecture
│   ├── adrs/
│   ├── integrations/
│   ├── domain/
│   ├── c4-diagrams/
│   └── academic-patterns/
│
├── research/                    # Stage 3: Integrated research
│   ├── business/
│   ├── market-research/
│   ├── domain/
│   ├── infrastructure/
│   ├── source-materials/
│   └── archive/                 # Stage 4: Historical
│
└── process/                     # Stage 3: Internal processes
    └── {topic}/

4.2 Analysis Directory Structure

Pattern: internal/analysis/{category}/{topic}/

Files (Required):

{topic}-assessment-YYYY-MM-DD.md — Primary assessment document
MANIFEST.md — Research lineage and artifacts tracking
README.md — Directory navigation

Files (Optional):

executive-summary.md — Promoted from pipeline
recommendations.md — Actionable next steps
integration-status.md — Tracking integration progress
key-findings.md — Summary of critical insights

Example:

internal/analysis/technology-evaluation/agent-labs/
├── agent-labs-scaling-assessment-2026-02-16.md
├── executive-summary.md
├── recommendations.md
├── MANIFEST.md
└── README.md

4.3 Architecture Integration Structure

SDDs/TDDs for Integrations:

internal/architecture/integrations/{topic}/
├── SDD-{topic}-integration.md
├── TDD-{topic}-integration.md
├── c4-diagrams/
│   ├── context.md
│   ├── container.md
│   └── component.md
└── README.md

SDDs/TDDs for Domain Features:

internal/architecture/domain/{industry}/
├── SDD-{industry}-domain-features.md
├── TDD-{industry}-data-models.md
├── workflows/
└── README.md

ADRs:

internal/architecture/adrs/
├── ADR-XXX-{topic}-adoption.md              # Integration decisions
├── ADR-XXX-{topic}-rejection.md             # No-Go decisions
└── rejected/
    └── ADR-XXX-{topic}-superseded.md        # Reversed decisions

Academic Patterns:

internal/architecture/academic-patterns/
├── {paper-topic}-implementation-guide.md
├── {paper-topic}-coditect-adaptation.md
└── README.md

4.4 Research Integration Structure

Business Research:

internal/research/business/
├── cases/
│   └── {topic}/
│       ├── business-case.md
│       ├── executive-summary.md
│       └── MANIFEST.md
├── models/
│   └── {topic}/
│       ├── {topic}_v{X}.json
│       ├── {topic}_v{X}.xlsx
│       └── assumptions.md
└── executive-briefs/
    └── {topic}-brief-YYYY-MM-DD.md

Market Research:

internal/research/market-research/
├── competitors/
│   └── {company}/
│       ├── competitive-analysis-YYYY-MM-DD.md
│       ├── feature-comparison.md
│       └── MANIFEST.md
├── competitive-analysis/
│   ├── feature-matrix-YYYY-MM-DD.md
│   └── positioning-strategy-YYYY-MM-DD.md
├── sizing/
│   ├── TAM-SAM-SOM-analysis-YYYY-MM-DD.md
│   └── market-opportunity-assessment-YYYY-MM-DD.md
└── dashboards/
    └── {topic}-competitive-comparison.jsx

Domain Research:

internal/research/domain/{industry}/
├── glossary.md
├── workflows/
│   ├── {workflow-name}.md
│   └── {workflow-name}.mermaid
├── compliance/
│   ├── requirements-checklist.md
│   └── regulatory-framework.md
└── papers/
    ├── {paper-id}.pdf
    ├── {paper-id}.md
    └── {paper-id}.udom.json

Source Materials:

internal/research/source-materials/
├── papers/
│   └── {research-topic}/
│       ├── {paper-id}.pdf
│       ├── {paper-id}.md
│       └── {paper-id}.udom.json
├── transcripts/
│   └── {source}/
│       └── {episode-title}.txt
└── README.md

4.5 Process Documentation Structure

internal/process/{topic}/
├── {topic}-process-guide.md
├── {topic}-workflow.mermaid
├── {topic}-templates/
└── README.md

4.6 Dashboard Lifecycle

Staging Location:

analyze-new-artifacts/{topic}/dashboards/
├── tech-architecture-analyzer.jsx
├── strategic-fit-dashboard.jsx
├── coditect-integration-playbook.jsx
└── executive-decision-brief.jsx

Promotion Decision:

Dashboard Type	Promotion Path	Criteria
Tech Architecture Analyzer	Rarely promoted	Research-specific; stays in staging
Strategic Fit Dashboard	Rarely promoted	Research-specific; stays in staging
Integration Playbook	`internal/research/market-research/dashboards/`	If reusable for future integrations
Executive Decision Brief	`internal/research/business/executive-briefs/`	If board-level artifact
Competitive Comparison	`internal/research/market-research/dashboards/`	If reusable for competitor tracking
Implementation Planner	Rarely promoted	Project-specific; stays in staging

Rationale: JSX dashboards are often research-specific and don't need permanent storage. Only promote if:

Reusable across multiple research topics
Part of executive reporting framework
Adopted for production use in CODITECT UI

5. Manifest System

5.1 Purpose

The manifest system provides:

Lineage tracking — What research produced what outputs
Artifact inventory — Complete list of generated artifacts
Promotion history — Where artifacts moved and when
Source attribution — Original materials referenced
Integration status — Which artifacts were integrated into CODITECT

5.2 Manifest Format

File: MANIFEST.md (one per research topic)

Location: Co-located with assessment document in internal/analysis/{category}/{topic}/

Template:

---
title: "Research Manifest: {Topic}"
research_date: 'YYYY-MM-DD'
research_category: [technology-evaluation|academic|competitive-intelligence|business-market|domain|process-internal]
pipeline_version: [manual|adr-206-v1.0]
status: [active|integrated|archived]
integration_status: [pending|partial|complete|rejected]
---

# Research Manifest: {Topic}

## Metadata

| Field | Value |
|-------|-------|
| **Research Date** | YYYY-MM-DD |
| **Researcher** | Claude (Model) / Human Name |
| **Category** | {Category} |
| **Pipeline** | [Manual / ADR-206 v1.0] |
| **Staging Location** | `analyze-new-artifacts/{topic}/` |
| **Analysis Location** | `internal/analysis/{category}/{topic}/` |
| **Status** | [Active / Integrated / Archived] |

---

## Research Objective

[1-2 sentences: What question was this research trying to answer?]

---

## Source Materials

| Type | Source | Location |
|------|--------|----------|
| PDF Paper | {Title} | `{path or URL}` |
| Transcript | {YouTube/Podcast} | `{path or URL}` |
| GitHub Repo | {Repo Name} | `{URL}` |
| Web Article | {Title} | `{URL}` |

---

## Generated Artifacts

### Phase 1: Research Artifacts

| Artifact | Staging Path | Promoted To | Status |
|----------|-------------|-------------|--------|
| Executive Summary | `staging/{topic}/executive-summary.md` | `analysis/{category}/{topic}/` | ✅ Promoted |
| Assessment Document | `staging/{topic}/assessment.md` | `analysis/{category}/{topic}/` | ✅ Promoted |
| SDD | `staging/{topic}/sdd.md` | `architecture/integrations/{topic}/` | ✅ Integrated |
| TDD | `staging/{topic}/tdd.md` | `architecture/integrations/{topic}/` | ✅ Integrated |
| C4 Architecture | `staging/{topic}/c4-architecture.md` | `architecture/c4-diagrams/integrations/{topic}/` | ✅ Integrated |
| Glossary | `staging/{topic}/glossary.md` | — | ❌ Not promoted |
| Mermaid Diagrams | `staging/{topic}/mermaid-diagrams.md` | — | ❌ Not promoted |

### Phase 2: Dashboards

| Dashboard | Staging Path | Promoted To | Status |
|-----------|-------------|-------------|--------|
| Tech Architecture Analyzer | `staging/{topic}/dashboards/tech-architecture-analyzer.jsx` | — | ❌ Not promoted |
| Strategic Fit Dashboard | `staging/{topic}/dashboards/strategic-fit-dashboard.jsx` | — | ❌ Not promoted |
| Integration Playbook | `staging/{topic}/dashboards/coditect-integration-playbook.jsx` | `research/market-research/dashboards/` | ✅ Promoted |
| Executive Decision Brief | `staging/{topic}/dashboards/executive-decision-brief.jsx` | — | ❌ Not promoted |

### Phase 3: ADRs

| ADR | Staging Path | Promoted To | Status |
|-----|-------------|-------------|--------|
| ADR-XXX-{topic}-adoption | `staging/{topic}/adrs/ADR-001-*.md` | `architecture/adrs/ADR-XXX-{topic}-adoption.md` | ✅ Integrated |
| ADR-XXX-{topic}-integration-pattern | `staging/{topic}/adrs/ADR-002-*.md` | `architecture/adrs/ADR-XXX-{topic}-integration-pattern.md` | ✅ Integrated |

---

## Integration Summary

### Decisions Made

- **Go/No-Go:** [Go | No-Go | Conditional]
- **Decision ADR:** [ADR-XXX reference]
- **Decision Date:** YYYY-MM-DD

### Integrated Components

- ✅ SDD integrated into `internal/architecture/integrations/{topic}/`
- ✅ TDD integrated into `internal/architecture/integrations/{topic}/`
- ✅ 2 ADRs added to architecture decision log
- ❌ Dashboards not adopted (research-specific)

### Remaining in Staging

- Glossary (topic-specific, not reusable)
- Mermaid diagrams (duplicates C4 diagrams)
- 4 dashboards (research-specific, not production)

---

## Follow-up Actions

- [ ] Complete implementation per SDD/TDD (Track: X, Task: X.X.X)
- [ ] Update integration status when implementation complete
- [ ] Archive staging directory after 90 days (2026-05-XX)

---

## References

- **Analysis Document:** `{path}`
- **ADRs:** `internal/architecture/adrs/ADR-XXX-*.md`
- **Integration Location:** `internal/architecture/integrations/{topic}/`

---

**Manifest Version:** 1.0.0
**Last Updated:** YYYY-MM-DD
**Maintainer:** [Agent/Human Name]

5.3 Manifest Automation

ADR-206 Pipeline Integration:

The research pipeline orchestrator (ADR-206) auto-generates manifests on completion:

def generate_manifest(pipeline_result: PipelineResult, config: PipelineConfig):
    """Auto-generate manifest from pipeline execution."""
    manifest = {
        "research_date": datetime.now().isoformat(),
        "researcher": "Claude (Opus 4.6)",  # or model used
        "category": config.category,
        "pipeline": "ADR-206 v1.0",
        "staging_location": config.output_dir,
        "source_materials": config.sources,
        "artifacts": {
            "phase1": pipeline_result.research_artifacts,
            "phase2": pipeline_result.dashboards,
            "phase3": pipeline_result.prompts,
        },
        "status": "active",
        "integration_status": "pending",
    }
    write_manifest(manifest, config.output_dir / "MANIFEST.md")

Manual Research:

For manual research sessions (not using ADR-206 pipeline), use command:

/generate-manifest --topic "{topic}" --category "{category}"

Interactive prompts collect:

Source materials
Generated artifacts (manual listing or auto-scan staging dir)
Research objective

5.4 Master Manifest Index

Location: internal/research/RESEARCH-MANIFEST-INDEX.md

Purpose: Searchable catalog of all research with status tracking.

Format:

# CODITECT Research Manifest Index

**Last Updated:** YYYY-MM-DD
**Total Research Topics:** XXX
**Active:** XX | **Integrated:** XX | **Archived:** XX

---

## Technology Evaluation

| Topic | Date | Status | Integration | Analysis Path | Manifest |
|-------|------|--------|-------------|---------------|----------|
| agent-labs | 2026-02-16 | Integrated | Complete | `analysis/technology-evaluation/agent-labs/` | [MANIFEST](../analysis/technology-evaluation/agent-labs/MANIFEST.md) |
| copilotkit | 2026-01-22 | Active | Partial | `analysis/technology-evaluation/copilotkit/` | [MANIFEST](../analysis/technology-evaluation/copilotkit/MANIFEST.md) |

## Academic Research

| Topic | Date | Status | Integration | Analysis Path | Manifest |
|-------|------|--------|-------------|---------------|----------|
| scaling-agent-systems | 2026-02-16 | Integrated | Complete | `analysis/academic/scaling-agent-systems/` | [MANIFEST](../analysis/academic/scaling-agent-systems/MANIFEST.md) |

[... continues for all 6 categories ...]

Automation:

# Auto-update master index
python3 scripts/research/update-manifest-index.py

# Search manifests
python3 scripts/research/search-manifests.py --keyword "multi-agent"

6. Promotion Criteria & Decision Matrix

6.1 Promotion Decision Tree

Research Complete in Staging
    ↓
    ├─→ Assessment Document Written? ────→ YES ──→ Promote to Analysis
    │                                       ↓
    │                                    NO ──→ Research incomplete, stay in staging
    ↓
Analysis Complete
    ↓
    ├─→ Go Decision Made? ────→ YES ──→ Promote SDDs/TDDs/ADRs to Integration
    │                            ↓
    │                         NO ──→ Stay in Analysis (assessment only)
    ↓
Time-Sensitive Research?
    ↓
    ├─→ Age > 12 months? ────→ YES ──→ Archive (if not refreshed)
    │                            ↓
    │                         NO ──→ Stay in Analysis

6.2 Promotion Criteria by Artifact Type

Executive Summary

Criteria:

✅ 1-2 pages (500-1000 words)
✅ Contains: objective, findings, recommendation, impact
✅ Standalone (readable without other artifacts)
✅ No placeholder content

Promotion Path: Staging → Analysis (always)

Location: internal/analysis/{category}/{topic}/executive-summary.md

Assessment Document

Criteria:

✅ 5-15 pages (2000-8000 words)
✅ Synthesizes research findings (not just pipeline output)
✅ Contains analysis, not just summaries
✅ References source materials
✅ Includes recommendations with rationale
✅ Quality validated (no broken links, no TODO markers)

Promotion Path: Staging → Analysis (always, if research complete)

Location: internal/analysis/{category}/{topic}/{topic}-assessment-YYYY-MM-DD.md

SDD (Software Design Document)

Criteria:

✅ Follows CODITECT SDD template
✅ Contains C4 context/container/component views
✅ Describes integration architecture
✅ Referenced by ADR with Go decision
✅ Reviewed for technical accuracy

Promotion Path:

Staging → Analysis (if referenced by assessment)
Analysis → Integration (if Go decision)

Location (Analysis): internal/analysis/{category}/{topic}/sdd.md Location (Integration): internal/architecture/integrations/{topic}/SDD-{topic}-integration.md

TDD (Technical Design Document)

Criteria:

✅ Follows CODITECT TDD template
✅ Contains data models, API specs, sequence diagrams
✅ Describes integration technical details
✅ Referenced by ADR with Go decision
✅ Reviewed for technical accuracy

Promotion Path:

Staging → Analysis (if referenced by assessment)
Analysis → Integration (if Go decision)

Location (Analysis): internal/analysis/{category}/{topic}/tdd.md Location (Integration): internal/architecture/integrations/{topic}/TDD-{topic}-integration.md

ADR (Architecture Decision Record)

Criteria:

✅ Follows ADR template (Context, Decision, Consequences, Alternatives)
✅ Status finalized (Proposed → Accepted | Rejected)
✅ Numbered (ADR-XXX assigned)
✅ Indexed in ADR index
✅ Cross-referenced by assessment document

Promotion Path:

Staging → Analysis (draft ADRs, if referenced)
Analysis → Integration (when status finalized)

Location (Analysis): internal/analysis/{category}/{topic}/adrs/ADR-draft-*.md Location (Integration - Accepted): internal/architecture/adrs/ADR-XXX-{topic}.md Location (Integration - Rejected): internal/architecture/adrs/rejected/ADR-XXX-{topic}-rejected.md

Business Case

Criteria:

✅ Contains: market opportunity, financial projections, ROI, risks
✅ Financial model attached (Excel/JSON)
✅ Executive summary (1-page)
✅ Board-ready (no draft content)

Promotion Path: Staging → Analysis → Integration (always, if complete)

Location (Analysis): internal/analysis/business-market/{topic}/business-case.md Location (Integration): internal/research/business/cases/{topic}/business-case.md

Glossary

Criteria:

✅ Domain-specific terminology (10+ terms)
✅ Definitions accurate and sourced
✅ Reusable for future research in same domain

Promotion Path: Staging → Integration (only if reusable)

Location (Integration): internal/research/domain/{industry}/glossary.md

Note: Most glossaries stay in staging (research-specific, not reusable).

Dashboards (JSX)

Criteria:

✅ Reusable across multiple research topics
✅ Part of executive reporting framework
✅ OR: Adopted for production use in CODITECT UI

Promotion Path: Staging → Integration (rarely)

Location (Integration): internal/research/market-research/dashboards/{dashboard-name}.jsx

Note: Most dashboards stay in staging (research-specific).

Source Materials (PDFs, Transcripts)

Criteria:

✅ Directly referenced by analysis document
✅ Not easily regenerable (e.g., proprietary paper, internal transcript)

Promotion Path: Staging → Integration (rarely)

Location (Integration): internal/research/source-materials/{type}/{topic}/

Note: Most source materials stay in staging and referenced by URL/path. Only promote if:

Proprietary/paywalled (can't re-access)
Internal materials (transcripts of internal meetings)
Critical reference for understanding analysis

6.3 Promotion Workflow

Automated (ADR-206 Pipeline):

async def promote_artifacts(pipeline_result: PipelineResult, config: PipelineConfig):
    """Auto-promote artifacts based on pipeline completion."""

    # Stage 1: Always promote to Analysis
    promote_to_analysis(
        artifacts=["executive-summary.md", "assessment.md"],
        destination=f"internal/analysis/{config.category}/{config.topic}/"
    )

    # Stage 2: Conditional promotion to Integration
    if config.genesis_handoff:  # --genesis flag set
        # Go decision implied by handoff to /new-project
        promote_to_integration(
            artifacts=["sdd.md", "tdd.md", "c4-architecture.md", "adrs/"],
            destination=f"internal/architecture/integrations/{config.topic}/"
        )
    else:
        # Manual promotion after decision
        log_promotion_pending(artifacts=["sdd.md", "tdd.md", "adrs/"])

Manual (Post-Research):

# Promote assessment to analysis
/research-promote --topic "{topic}" --to analysis

# Promote SDDs/TDDs/ADRs to integration (after Go decision)
/research-promote --topic "{topic}" --to integration --artifacts sdd,tdd,adrs

# Archive old research
/research-archive --topic "{topic}" --reason "superseded by {new-topic}"

7. Naming Conventions

7.1 Directory Naming

Pattern: {category}-{topic} (kebab-case, lowercase)

Examples:

internal/analysis/technology-evaluation/agent-labs/
internal/analysis/academic/scaling-agent-systems/
internal/analysis/competitive-intelligence/palantir/
internal/analysis/business-market/product-market-fit/
internal/analysis/domain/bioscience-workorders/
internal/analysis/process-internal/system-prompt/

Rules:

Use kebab-case (hyphens, not underscores or camelCase)
Lowercase only (no UPPERCASE)
No coditect- prefix (redundant in CODITECT repo)
No version numbers in directory names (use file versioning)
No dates in directory names (use file naming for dates)

7.2 File Naming

Assessment Documents:

{topic}-assessment-YYYY-MM-DD.md

Executive Summaries:

executive-summary.md  (standard name, no date)

SDDs/TDDs:

SDD-{topic}-integration.md
TDD-{topic}-integration.md
SDD-{industry}-domain-features.md
TDD-{industry}-data-models.md

ADRs:

ADR-XXX-{topic}-{decision-summary}.md

Business Cases:

{topic}-business-case-YYYY-MM-DD.md

Financial Models:

{topic}_v{X}.{Y}.json
{topic}_v{X}.{Y}.xlsx

Dashboards:

{dashboard-name}.jsx  (kebab-case)

Manifests:

MANIFEST.md  (standard name, all research topics)

7.3 Frontmatter Standards

All Documents (Required Fields):

---
title: "Document Title"
type: [assessment|sdd|tdd|adr|business-case|reference]
audience: contributor
status: [draft|active|integrated|archived]
created: 'YYYY-MM-DD'
updated: 'YYYY-MM-DD'
summary: "One-line AI agent summary"
tags:
- research
- {category}
- {topic}
---

Research-Specific Fields:

research_date: 'YYYY-MM-DD'
research_category: [technology-evaluation|academic|competitive-intelligence|business-market|domain|process-internal]
pipeline_version: [manual|adr-206-v1.0]
integration_status: [pending|partial|complete|rejected]
superseded_by: 'path/to/newer-doc.md'  # If archived
archive_reason: 'Brief explanation'     # If archived

8. Integration with ADR-206 Pipeline

8.1 ADR-206 Pipeline Outputs

Current Output (ADR-206):

{output-dir}/
├── README.md
├── research-context.json
├── 1-2-3-detailed-quick-start.md
├── coditect-impact.md
├── executive-summary.md
├── sdd.md
├── tdd.md
├── c4-architecture.md
├── glossary.md
├── mermaid-diagrams.md
├── follow-up-prompts.md
├── adrs/
│   ├── ADR-001-*.md
│   └── ADR-00N-*.md
├── dashboards/
│   ├── tech-architecture-analyzer.jsx
│   ├── strategic-fit-dashboard.jsx
│   ├── coditect-integration-playbook.jsx
│   └── executive-decision-brief.jsx
└── pipeline-report.json

Enhanced Output (with this proposal):

{output-dir}/
├── README.md
├── MANIFEST.md                          # ⬅️ NEW: Auto-generated manifest
├── research-context.json
├── 1-2-3-detailed-quick-start.md
├── coditect-impact.md
├── executive-summary.md
├── sdd.md
├── tdd.md
├── c4-architecture.md
├── glossary.md
├── mermaid-diagrams.md
├── follow-up-prompts.md
├── adrs/
│   ├── ADR-001-*.md
│   └── ADR-00N-*.md
├── dashboards/
│   ├── tech-architecture-analyzer.jsx
│   ├── strategic-fit-dashboard.jsx
│   ├── coditect-integration-playbook.jsx
│   └── executive-decision-brief.jsx
├── pipeline-report.json
└── PROMOTION-CHECKLIST.md               # ⬅️ NEW: Next steps for user

8.2 Pipeline Enhancement: Auto-Promotion

New Pipeline Phase 4.5: Artifact Promotion (Optional)

# Phase 4: Genesis (optional) — EXISTING
if config.genesis:
    await self.handoff_to_new_project(artifacts, dashboards, prompts)

# Phase 4.5: Promotion (optional) — NEW
if config.auto_promote:
    await self.promote_to_analysis(artifacts, config)

    if config.genesis:  # If genesis flag set, Go decision implied
        await self.promote_to_integration(artifacts, config)

New CLI Flags:

/research-pipeline "{topic}" \
  --urls https://example.com \
  --github https://github.com/org/repo \
  --genesis \                # Handoff to /new-project (Go decision)
  --auto-promote             # Auto-promote to analysis + integration

Promotion Logic:

async def promote_to_analysis(artifacts: dict, config: PipelineConfig):
    """Auto-promote core artifacts to analysis."""
    analysis_dir = Path(f"internal/analysis/{config.category}/{config.topic}")
    analysis_dir.mkdir(parents=True, exist_ok=True)

    # Always promote these to analysis
    promote_files = [
        "executive-summary.md",
        "coditect-impact.md",  # Rename to assessment
    ]

    for file in promote_files:
        src = Path(config.output_dir) / file
        dst = analysis_dir / file
        shutil.copy2(src, dst)

    # Create README and MANIFEST
    create_analysis_readme(analysis_dir, config)
    create_manifest(analysis_dir, artifacts, config)

async def promote_to_integration(artifacts: dict, config: PipelineConfig):
    """Auto-promote SDDs/TDDs/ADRs to integration (if Go decision)."""
    integration_dir = Path(f"internal/architecture/integrations/{config.topic}")
    integration_dir.mkdir(parents=True, exist_ok=True)

    # Promote SDDs/TDDs
    shutil.copy2(
        Path(config.output_dir) / "sdd.md",
        integration_dir / f"SDD-{config.topic}-integration.md"
    )
    shutil.copy2(
        Path(config.output_dir) / "tdd.md",
        integration_dir / f"TDD-{config.topic}-integration.md"
    )

    # Promote ADRs (renumber with next available ADR-XXX)
    for adr_draft in (Path(config.output_dir) / "adrs").glob("*.md"):
        adr_num = get_next_adr_number()
        adr_name = adr_draft.stem.replace("ADR-001", f"ADR-{adr_num:03d}")
        shutil.copy2(adr_draft, Path("internal/architecture/adrs") / f"{adr_name}.md")

8.3 Pipeline Enhancement: Category Auto-Detection

Problem: User must specify --category flag manually.

Solution: Auto-detect category based on research inputs.

def auto_detect_category(config: PipelineConfig) -> str:
    """Auto-detect research category from inputs."""

    # GitHub repo URL → technology evaluation
    if config.github:
        return "technology-evaluation"

    # arXiv paper URL → academic
    if any("arxiv.org" in url for url in config.urls):
        return "academic"

    # Competitor domain → competitive intelligence
    competitor_domains = ["anthropic.com", "openai.com", "microsoft.com"]
    if any(domain in url for url in config.urls for domain in competitor_domains):
        return "competitive-intelligence"

    # Financial model keywords → business-market
    if "financial" in config.topic.lower() or "market" in config.topic.lower():
        return "business-market"

    # Default: prompt user
    return ask_user_category()

9. Migration Plan for Existing 98 Directories

9.1 Migration Strategy

Approach: Semi-automated with human review.

Phases:

Categorize — Auto-assign categories to 98 directories
Assess — Human review: promote vs. delete vs. archive
Promote — Move valuable artifacts to permanent locations
Archive — Move historical research to archive
Delete — Remove staging directories (after 90-day grace period)

9.2 Phase 1: Categorization (Automated)

Script: scripts/research/categorize-staging.py

Algorithm:

def categorize_directory(dir_name: str, contents: list) -> str:
    """Auto-assign category based on directory name and contents."""

    # Technology evaluation patterns
    tech_keywords = ["api", "tool", "framework", "sdk", "integration", "plugin"]
    if any(kw in dir_name.lower() for kw in tech_keywords):
        return "technology-evaluation"

    # Academic research patterns
    if "research" in dir_name and any(ext in contents for ext in [".pdf", ".tex"]):
        return "academic"

    # Competitive intelligence patterns
    competitor_names = ["anthropic", "openai", "microsoft", "palantir", "kimi"]
    if any(comp in dir_name.lower() for comp in competitor_names):
        return "competitive-intelligence"

    # Business patterns
    business_keywords = ["financial", "market", "business", "runway", "seed"]
    if any(kw in dir_name.lower() for kw in business_keywords):
        return "business-market"

    # Domain patterns
    domain_keywords = ["healthcare", "biotech", "legal", "regulatory", "compliance"]
    if any(kw in dir_name.lower() for kw in domain_keywords):
        return "domain"

    # Process patterns
    process_keywords = ["system-prompt", "docker", "installation", "process"]
    if any(kw in dir_name.lower() for kw in process_keywords):
        return "process-internal"

    # Default: manual review needed
    return "uncategorized"

Output: staging-categorization-report.json

{
  "categorized": {
    "technology-evaluation": 24,
    "academic": 18,
    "competitive-intelligence": 15,
    "business-market": 12,
    "domain": 11,
    "process-internal": 8
  },
  "uncategorized": 10,
  "total": 98,
  "details": [
    {
      "directory": "coditect-agent-labs-research",
      "category": "technology-evaluation",
      "confidence": 0.95,
      "rationale": "GitHub repo URL present, tool evaluation artifacts"
    },
    ...
  ]
}

9.3 Phase 2: Assessment (Human Review)

For each categorized directory, decide:

Decision	Action	Criteria
Promote	Move to `internal/analysis/{category}/`	Valuable findings, reusable
Archive	Move to `internal/research/archive/{category}/`	Historical value, no current relevance
Delete	Remove from staging (after grace period)	No value, regenerable, duplicates

Review Tool:

/research-review --category technology-evaluation

# Interactive prompts:
# - Show directory contents
# - Show any existing assessment docs
# - Suggest promotion path
# - User decides: promote / archive / delete

9.4 Phase 3: Promotion (Semi-Automated)

For directories marked "promote":

/research-promote \
  --source "analyze-new-artifacts/{topic}" \
  --category "{category}" \
  --artifacts "executive-summary.md,sdd.md,tdd.md,adrs/"

# Actions:
# 1. Create internal/analysis/{category}/{topic}/
# 2. Copy specified artifacts
# 3. Generate MANIFEST.md
# 4. Create README.md
# 5. Update RESEARCH-MANIFEST-INDEX.md
# 6. Add to git

Validation:

Check frontmatter present and valid
Verify no broken links
Confirm no TODO/FIXME markers
Lint markdown

9.5 Phase 4: Archive (Manual)

For directories marked "archive":

/research-archive \
  --source "analyze-new-artifacts/{topic}" \
  --category "{category}" \
  --reason "Superseded by {new-topic}"

# Actions:
# 1. Create internal/research/archive/{category}/{topic}/
# 2. Copy all artifacts
# 3. Add ARCHIVE-NOTICE.md with reason
# 4. Update MANIFEST.md with archive status
# 5. Update RESEARCH-MANIFEST-INDEX.md
# 6. Add to git

9.6 Phase 5: Deletion (Automated with Grace Period)

For directories marked "delete":

# Mark for deletion (grace period: 90 days)
/research-mark-delete --topic "{topic}" --reason "No value, regenerable"

# Moves to internal grace period directory
mv analyze-new-artifacts/{topic} analyze-new-artifacts/.pending-deletion/{topic}

# After 90 days, auto-delete (cron job)
# scripts/research/cleanup-expired-deletions.sh

Grace Period Purpose:

Allow recovery if deletion was mistake
User can un-mark: /research-unmark-delete --topic "{topic}"

9.7 Migration Timeline

Phase	Duration	Effort	Status
1. Categorization	1 day	Automated script	Not started
2. Assessment	5 days	Human review (98 dirs × 20 min avg)	Not started
3. Promotion	2 days	Semi-automated (30 dirs × 15 min)	Not started
4. Archive	1 day	Manual (20 dirs × 10 min)	Not started
5. Deletion	90 days	Automated (grace period)	Not started
Total	99 days	~40 hours human effort	Not started

10. Comparison of Alternatives

10.1 Alternative 1: Keep Current Ad-Hoc System

Description: Continue using analyze-new-artifacts/ as permanent storage, no organization.

Pros:

Zero migration effort
No new tooling needed

Cons:

❌ No searchability (6.1GB of unsearchable files)
❌ No lifecycle management (files never deleted)
❌ No integration with ADR-206 pipeline
❌ No distinction between staging and permanent
❌ Duplicates and inconsistency continue

Verdict: ❌ Rejected — Scales poorly, loses knowledge.

10.2 Alternative 2: Git-Track Staging Area

Description: Remove .gitignore from analyze-new-artifacts/, track everything.

Pros:

All research preserved in git history
No promotion workflow needed

Cons:

❌ 6.1GB in git (PDFs, large artifacts)
❌ No distinction between staging and permanent
❌ Clutters git history with work-in-progress
❌ Slows down clones/pulls

Verdict: ❌ Rejected — Git is not a storage system for ephemeral files.

10.3 Alternative 3: External Knowledge Base (Notion, Confluence)

Description: Store research in external wiki/knowledge base, reference from CODITECT.

Pros:

Better search than git
Rich media support (embeds, images)
Collaboration features

Cons:

❌ Separates research from code (no single source of truth)
❌ Requires external service (vendor lock-in)
❌ No integration with ADR-206 pipeline
❌ Markdown export/import overhead

Verdict: ❌ Rejected — CODITECT is markdown-first; keep research co-located with architecture.

10.4 Alternative 4: Database-Backed Knowledge Graph

Description: Store research in SQLite/PostgreSQL with semantic search and graph relationships.

Pros:

Best searchability (full-text search, semantic vectors)
Rich relationships (research → ADRs → code)
Query flexibility

Cons:

❌ Over-engineered for current scale (~100 research topics)
❌ Requires migration from markdown to database
❌ Loses markdown-first simplicity
❌ Requires additional tooling/maintenance

Verdict: ❌ Rejected — Overkill for current needs. Markdown + manifests + grep is sufficient.

10.5 Alternative 5: This Proposal (Markdown + Manifests)

Description: Markdown files in git, organized by category, tracked by manifests, lifecycle-managed.

Pros:

✅ Markdown-first (CODITECT standard)
✅ Git-tracked (version history)
✅ Searchable (grep, semantic search on markdown)
✅ Scalable (can handle 1000+ research topics)
✅ Integrates with ADR-206 pipeline
✅ Clear lifecycle (staging → analysis → integration → archive)

Cons:

Requires migration effort (40 hours)
Requires new tooling (promotion scripts, review commands)

Verdict: ✅ RECOMMENDED — Best balance of simplicity, scalability, and integration.

11. Success Metrics

11.1 Quantitative Metrics

Metric	Baseline (Now)	Target (Post-Migration)
Staging Clutter	98 directories	≤13 active research topics
Searchable Research	20% (analysis only)	100% (all permanent)
Promotion Time	Manual (hours)	Automated (seconds)
Artifact Loss Rate	80% (only assessment promoted)	0% (systematic promotion)
Research Findability	Low (no index)	High (manifest index)
Git Repo Size	+6.1GB (if tracked)	+200MB (selective promotion)

11.2 Qualitative Metrics

Metric	Success Criteria
Discoverability	Any contributor can find research via manifest index
Reusability	SDDs/TDDs/ADRs reused across projects
Integration	ADR-206 pipeline outputs directly to permanent locations
Lifecycle Clarity	Clear rules for promotion/archival (no ambiguity)
Maintainability	Manifest system self-documenting, low maintenance

12. Implementation Roadmap

12.1 Phase 1: Foundation (Week 1)

Tasks:

Create directory structure (internal/analysis/{6 categories}/)
Create manifest template (MANIFEST.md)
Create master manifest index (RESEARCH-MANIFEST-INDEX.md)
Write promotion scripts (/research-promote, /research-archive)
Write categorization script (scripts/research/categorize-staging.py)

Deliverables:

Empty directory structure ready for migration
Manifest system operational
Automation scripts tested

12.2 Phase 2: Pilot Migration (Week 2)

Tasks:

Categorize 10 sample directories (2 per category)
Manually promote 5 directories to test workflow
Archive 3 directories to test archive workflow
Delete 2 directories to test deletion workflow
Iterate on tooling based on learnings

Deliverables:

10 research topics migrated and validated
Workflow refined and documented
Migration guide for remaining 88 directories

12.3 Phase 3: Bulk Migration (Week 3-4)

Tasks:

Auto-categorize remaining 88 directories
Human review: assess each directory (promote/archive/delete)
Batch promote 25 directories
Batch archive 15 directories
Mark 48 directories for deletion (grace period)

Deliverables:

40 research topics migrated (30 promoted + 15 archived)
48 directories marked for deletion
Manifest index populated

12.4 Phase 4: ADR-206 Integration (Week 5)

Tasks:

Enhance ADR-206 pipeline with manifest generation
Add --auto-promote flag to pipeline
Add --category auto-detection
Test full pipeline: research → staging → analysis → integration
Document new workflow in ADR-206

Deliverables:

ADR-206 pipeline fully integrated with promotion system
End-to-end automation: research → permanent storage
Documentation updated

12.5 Phase 5: Grace Period & Cleanup (Week 6-18)

Tasks:

Monitor 90-day grace period for marked deletions
Allow recovery of mis-marked directories
Auto-delete expired directories (after 90 days)
Final validation: all valuable research promoted

Deliverables:

Staging area cleaned (13 active topics only)
Zero artifact loss
System stable and maintainable

13. Risks & Mitigations

13.1 Risk: Accidental Deletion of Valuable Research

Likelihood: Medium Impact: High

Mitigation:

90-day grace period before deletion
Manual review required before marking for deletion
Recovery command: /research-unmark-delete
Manifest system tracks all artifacts (even deleted ones)

13.2 Risk: Inconsistent Categorization

Likelihood: Medium Impact: Medium

Mitigation:

Auto-categorization script with confidence scores
Human review of all categorizations
Re-categorization command: /research-recategorize
Category definitions clearly documented (Section 2)

13.3 Risk: Promotion Overhead Discourages Use

Likelihood: Medium Impact: High

Mitigation:

ADR-206 pipeline auto-promotes with --auto-promote flag
Semi-automated promotion (not fully manual)
Batch promotion for bulk operations
Clear promotion criteria (no ambiguity)

13.4 Risk: Manifest System Becomes Stale

Likelihood: Medium Impact: Medium

Mitigation:

Manifests auto-generated by ADR-206 pipeline
Validation checks on manifest updates
Master index auto-updated by scripts
Cron job: detect manifests needing updates

14. Appendices

14.1 Appendix A: Migration Script Pseudocode

def migrate_research_topic(topic: str, decision: str):
    """Migrate a single research topic."""

    if decision == "promote":
        # Step 1: Detect category
        category = auto_detect_category(topic)

        # Step 2: Create analysis directory
        analysis_dir = f"internal/analysis/{category}/{topic}"
        os.makedirs(analysis_dir, exist_ok=True)

        # Step 3: Copy artifacts
        artifacts_to_promote = [
            "executive-summary.md",
            "coditect-impact.md",
            "sdd.md",  # if exists
            "tdd.md",  # if exists
            "adrs/",   # if exists
        ]
        for artifact in artifacts_to_promote:
            if exists(f"analyze-new-artifacts/{topic}/{artifact}"):
                copy(f"analyze-new-artifacts/{topic}/{artifact}", analysis_dir)

        # Step 4: Generate manifest
        generate_manifest(analysis_dir, topic, category)

        # Step 5: Create README
        create_analysis_readme(analysis_dir, topic)

        # Step 6: Update master index
        update_manifest_index(topic, category, "active")

        # Step 7: Git add
        git_add(analysis_dir)

    elif decision == "archive":
        # Similar process, but to archive location
        pass

    elif decision == "delete":
        # Move to grace period directory
        os.rename(
            f"analyze-new-artifacts/{topic}",
            f"analyze-new-artifacts/.pending-deletion/{topic}"
        )
        create_deletion_notice(topic, grace_period_days=90)

14.2 Appendix B: Manifest Query Examples

# Find all research on multi-agent systems
grep -r "multi-agent" internal/analysis/*/*/MANIFEST.md

# Find all integrated research
grep -l "integration_status: complete" internal/analysis/*/*/MANIFEST.md

# Find all research from 2026-02
grep -l "research_date: '2026-02" internal/analysis/*/*/MANIFEST.md

# Find all ADR-206 pipeline research
grep -l "pipeline_version: adr-206" internal/analysis/*/*/MANIFEST.md

# Semantic search (if vector index available)
python3 scripts/research/semantic-search.py "How to scale agent orchestration?"

14.3 Appendix C: Example Promoted Directory

Scenario: Agent Labs research completed, Go decision made, artifacts promoted.

Before (Staging):

analyze-new-artifacts/coditect-agent-labs-research/
├── 2512.08296v2.pdf
├── I Built an Open-Source Rig That Measures Multi-Agent Architectures.md
├── artifacts/
│   ├── 1-2-3-detailed-quick-start.md
│   ├── coditect-impact.md
│   ├── executive-summary.md
│   ├── sdd.md
│   ├── tdd.md
│   ├── c4-architecture.md
│   ├── glossary.md
│   ├── mermaid-diagrams.md
│   ├── adrs/
│   │   ├── ADR-001-agent-labs-adoption.md
│   │   ├── ADR-002-integration-pattern.md
│   │   └── ADR-003-agent-orchestration-mapping.md
│   └── dashboards/
│       ├── tech-architecture-analyzer.jsx
│       ├── strategic-fit-dashboard.jsx
│       └── coditect-integration-playbook.jsx
└── README.md

After (Analysis):

internal/analysis/technology-evaluation/agent-labs/
├── agent-labs-scaling-assessment-2026-02-16.md  # Human-written synthesis
├── executive-summary.md                          # Promoted from staging
├── recommendations.md                            # Next steps
├── MANIFEST.md                                   # Research lineage
└── README.md                                     # Navigation

After (Integration):

internal/architecture/integrations/agent-labs/
├── SDD-agent-labs-integration.md
├── TDD-agent-labs-integration.md
├── c4-diagrams/
│   ├── context.md
│   ├── container.md
│   └── component.md
└── README.md

internal/architecture/adrs/
├── ADR-207-agent-labs-adoption.md                # Renumbered from ADR-001
├── ADR-208-agent-labs-integration-pattern.md     # Renumbered from ADR-002
└── ADR-209-agent-orchestration-mapping.md        # Renumbered from ADR-003

Staging (Remaining):

analyze-new-artifacts/coditect-agent-labs-research/
├── 2512.08296v2.pdf                              # Source material (stays)
├── I Built an Open-Source Rig....md              # Source material (stays)
├── artifacts/glossary.md                         # Not promoted (topic-specific)
├── artifacts/mermaid-diagrams.md                 # Not promoted (duplicates C4)
└── artifacts/dashboards/                         # Not promoted (research-specific)

15. Conclusion

This proposal establishes a comprehensive research artifact organization system for CODITECT with:

6-category taxonomy distinguishing technology evaluation, academic, competitive, business, domain, and process research
4-stage lifecycle providing clear progression from staging → analysis → integration → archive
Promotion criteria matrix defining exactly when/how artifacts move between stages
Permanent directory structure replacing ad-hoc organization with clear categorization
Manifest system tracking research lineage, artifacts, and integration status
ADR-206 integration enabling end-to-end automation from research → permanent storage

Impact:

87% reduction in staging clutter (98 → 13 directories)
Zero loss of valuable artifacts (systematic promotion)
100% searchability (manifest index + git-tracked)
End-to-end automation (ADR-206 pipeline → permanent locations)

Next Steps:

Review and approve this proposal
Begin Phase 1 implementation (directory structure + tooling)
Pilot migration on 10 sample directories
Bulk migration of remaining 88 directories
ADR-206 pipeline enhancement with auto-promotion

Timeline: 18 weeks (40 hours active work + 90-day grace period)

Document Status: Proposed — 2026-02-16 Author: Claude (Sonnet 4.5) Reviewers: TBD Approval Required: CODITECT Framework Architect, Research Team Lead

Version: 1.0.0 Last Updated: 2026-02-16 Compliance: CODITECT CLAUDE.md Standard v1.0.0

Executive Summary​

1. Problem Statement​

1.1 Current State Analysis​

1.2 ADR-206 Context​

2. Research Taxonomy (6 Categories)​

2.1 Taxonomy Design Principles​

2.2 Category Definitions​

Category 1: Technology Evaluation​

Category 2: Academic Research​

Category 3: Competitive Intelligence​

Category 4: Business & Market Research​

Category 5: Domain Research​

Category 6: Process & Internal Research​

2.3 Special Categories (Non-Research Content)​

Media & Transcripts​

UI/UX & Visual​

3. Artifact Lifecycle (4 Stages)​

3.1 Lifecycle Model​

3.2 Stage 1: STAGING​

3.3 Stage 2: ANALYSIS​

3.4 Stage 3: INTEGRATION​

3.5 Stage 4: ARCHIVE​

3.6 Lifecycle Decision Matrix​

4. Permanent Directory Structure​

4.1 Top-Level Organization​

4.2 Analysis Directory Structure​

4.3 Architecture Integration Structure​

4.4 Research Integration Structure​

4.5 Process Documentation Structure​

4.6 Dashboard Lifecycle​

5. Manifest System​

5.1 Purpose​

5.2 Manifest Format​

5.3 Manifest Automation​

5.4 Master Manifest Index​

6. Promotion Criteria & Decision Matrix​

6.1 Promotion Decision Tree​

6.2 Promotion Criteria by Artifact Type​

Executive Summary​

Assessment Document​

SDD (Software Design Document)​

TDD (Technical Design Document)​

ADR (Architecture Decision Record)​

Business Case​

Glossary​

Dashboards (JSX)​

Source Materials (PDFs, Transcripts)​

6.3 Promotion Workflow​

7. Naming Conventions​

7.1 Directory Naming​

7.2 File Naming​

7.3 Frontmatter Standards​

8. Integration with ADR-206 Pipeline​

8.1 ADR-206 Pipeline Outputs​

8.2 Pipeline Enhancement: Auto-Promotion​

8.3 Pipeline Enhancement: Category Auto-Detection​

9. Migration Plan for Existing 98 Directories​

9.1 Migration Strategy​

9.2 Phase 1: Categorization (Automated)​

9.3 Phase 2: Assessment (Human Review)​

9.4 Phase 3: Promotion (Semi-Automated)​

9.5 Phase 4: Archive (Manual)​

9.6 Phase 5: Deletion (Automated with Grace Period)​

9.7 Migration Timeline​

10. Comparison of Alternatives​

10.1 Alternative 1: Keep Current Ad-Hoc System​

10.2 Alternative 2: Git-Track Staging Area​

10.3 Alternative 3: External Knowledge Base (Notion, Confluence)​

10.4 Alternative 4: Database-Backed Knowledge Graph​

10.5 Alternative 5: This Proposal (Markdown + Manifests)​

11. Success Metrics​

11.1 Quantitative Metrics​

11.2 Qualitative Metrics​

12. Implementation Roadmap​

12.1 Phase 1: Foundation (Week 1)​

12.2 Phase 2: Pilot Migration (Week 2)​

12.3 Phase 3: Bulk Migration (Week 3-4)​

12.4 Phase 4: ADR-206 Integration (Week 5)​

12.5 Phase 5: Grace Period & Cleanup (Week 6-18)​

13. Risks & Mitigations​

Executive Summary

1. Problem Statement

1.1 Current State Analysis

1.2 ADR-206 Context

2. Research Taxonomy (6 Categories)

2.1 Taxonomy Design Principles

2.2 Category Definitions

Category 1: Technology Evaluation

Category 2: Academic Research

Category 3: Competitive Intelligence

Category 4: Business & Market Research

Category 5: Domain Research

Category 6: Process & Internal Research

2.3 Special Categories (Non-Research Content)

Media & Transcripts

UI/UX & Visual

3. Artifact Lifecycle (4 Stages)

3.1 Lifecycle Model

3.2 Stage 1: STAGING

3.3 Stage 2: ANALYSIS

3.4 Stage 3: INTEGRATION

3.5 Stage 4: ARCHIVE

3.6 Lifecycle Decision Matrix

4. Permanent Directory Structure

4.1 Top-Level Organization

4.2 Analysis Directory Structure

4.3 Architecture Integration Structure

4.4 Research Integration Structure

4.5 Process Documentation Structure

4.6 Dashboard Lifecycle

5. Manifest System

5.1 Purpose

5.2 Manifest Format

5.3 Manifest Automation

5.4 Master Manifest Index

6. Promotion Criteria & Decision Matrix

6.1 Promotion Decision Tree

6.2 Promotion Criteria by Artifact Type

Executive Summary

Assessment Document

SDD (Software Design Document)

TDD (Technical Design Document)

ADR (Architecture Decision Record)

Business Case

Glossary

Dashboards (JSX)

Source Materials (PDFs, Transcripts)

6.3 Promotion Workflow

7. Naming Conventions

7.1 Directory Naming

7.2 File Naming

7.3 Frontmatter Standards

8. Integration with ADR-206 Pipeline

8.1 ADR-206 Pipeline Outputs

8.2 Pipeline Enhancement: Auto-Promotion

8.3 Pipeline Enhancement: Category Auto-Detection

9. Migration Plan for Existing 98 Directories

9.1 Migration Strategy

9.2 Phase 1: Categorization (Automated)

9.3 Phase 2: Assessment (Human Review)

9.4 Phase 3: Promotion (Semi-Automated)

9.5 Phase 4: Archive (Manual)

9.6 Phase 5: Deletion (Automated with Grace Period)

9.7 Migration Timeline

10. Comparison of Alternatives

10.1 Alternative 1: Keep Current Ad-Hoc System

10.2 Alternative 2: Git-Track Staging Area

10.3 Alternative 3: External Knowledge Base (Notion, Confluence)

10.4 Alternative 4: Database-Backed Knowledge Graph

10.5 Alternative 5: This Proposal (Markdown + Manifests)

11. Success Metrics

11.1 Quantitative Metrics

11.2 Qualitative Metrics

12. Implementation Roadmap

12.1 Phase 1: Foundation (Week 1)

12.2 Phase 2: Pilot Migration (Week 2)

12.3 Phase 3: Bulk Migration (Week 3-4)

12.4 Phase 4: ADR-206 Integration (Week 5)

12.5 Phase 5: Grace Period & Cleanup (Week 6-18)

13. Risks & Mitigations