consolidate-research

What this skill does

# Consolidate Research

Synthesize research outputs from multiple sources into a unified, confidence-tiered, pattern-aware report with provenance tracking and quality metrics. Works standalone or manifest-driven from `create-research-brief`.

**Shared references**: `consolidation-manifest-schema.md` (schema), `pattern-registry.md` (patterns + defaults)

## Workflow Overview

```
1. RECEIVE INPUTS ─── Manifest detection, source identification
2. NORMALIZE INPUTS ── Model-specific extraction → common claim format
3. TRIAGE CLAIMS ───── Atomic decomposition, dependency graphs, provenance
4. RECONCILE ───────── Provenance-weighted disagreement, citation dedup
5. SYNTHESIZE ──────── Cross-domain synthesis, structural artifact merging
6. GENERATE OUTPUT ─── Pattern-specific template, quality metrics, freshness
7. SELF-REVIEW ─────── 7 mandatory checks before finalizing
```

---

## Step 1: Receive & Validate Inputs

### Manifest Detection

```
Input contains YAML block with `consolidation_manifest:` label?
│
├─ YES (manifest-driven)
│   ├─ Parse manifest fields: research_id, pattern, models, coverage_matrix
│   ├─ Extract consolidation config: mode, verification_priorities
│   ├─ Load pattern-specific output template from Pattern Registry
│   ├─ Note research chain context (upstream_id, inherited_constraints)
│   └─ Use model identifiers from manifest — do NOT guess sources
│
└─ NO (standalone mode)
    ├─ Ask for or infer: topic, sources, consolidation mode
    ├─ Derive pattern from content using Pattern Registry decision tree
    ├─ Use pattern defaults from Quick Reference table
    └─ Confirm inferred pattern + mode with user before proceeding
```

### Source Identification Heuristics

When manifest is absent and source is unlabeled, apply capability-aware signals:

| Signal | Likely Source | Strength |
|--------|--------------|----------|
| Deep reasoning chains, cross-domain analogies, self-corrections, hedged nuance | Claude Opus 4.6 | primary_researcher |
| Structured tables with dense citations, systematic catalogs, source appendix | Gemini 3.1 Pro Deep Research | structured_cataloger |
| Site-specific citations, temporal progression ("as of [date]"), intervention markers | GPT-5.2 Deep Research | targeted_investigator |
| Quick facts, recent dates, sentiment language, short-form responses | GPT-5.2 Chat | recency_validator |
| Proprietary data, specific methodologies, organizational context | Human analyst | domain_expert |

If uncertain after heuristic check, ask user to confirm. When manifest is present, always use `manifest.models[].model_id`.

### Validation

```
Research outputs provided?
├─ NO → Ask: "Please share the research outputs to consolidate."
└─ YES → Sources identifiable?
         ├─ NO → Ask: "Which source produced each output?"
         └─ YES → ≥2 sources?
                  ├─ NO → Warn: single-source consolidation has limited value.
                  │        Offer: proceed as structured review, or add sources.
                  └─ YES → Proceed to Step 2.
```

---

## Step 2: Normalize Inputs

Convert heterogeneous model outputs into a common claim format before triaging.

### Extraction Patterns by Source

| Source | Extract | Tag |
|--------|---------|-----|
| **Claude Opus 4.6** | Reasoning chains (analytical steps, not just conclusions) | `type: reasoning_chain` |
| | Conclusions linked to their supporting chains | `type: factual/causal` |
| | Web search findings with source URLs | `channel: web_search` |
| | Cross-domain insights and analogies | `type: cross_domain_synthesis` |
| | Self-review corrections (higher-confidence refinements) | `type: factual` |
| **Gemini 3.1 Pro** | Tables — preserve structure, do not flatten | `type: structural_artifact` |
| | Prose claims with inline citation mapping | `type: factual/causal` |
| | Source appendix — map citations to claims, compute quality scores | provenance metadata |
| | Comparison matrices with per-cell provenance | `type: structural_artifact` |
| | file_search-grounded claims | `channel: internal_document` |
| **GPT-5.2 Deep** | Assertions with temporal markers ("as of [date]") | `provenance.temporal_marker` |
| | Site-specific findings with source domain | `channel: site_restricted` |
| | Intervention-adjusted findings (higher targeted confidence) | boosted weight |
| | Timeline/progression narratives | `type: structural_artifact` |
| **GPT-5.2 Chat** | Quick facts with recency dates | `channel: quick_validation` |
| | Sentiment signals | `type: recommendation` |
| | *All Chat claims default to lower provenance weight* | validation role |

### Normalized Claim Format

```yaml
claim: {id: "C-{seq}", text: "...", type: factual|causal|quantitative|recommendation|unique|reasoning_chain|structural_artifact,
  source_model: claude-opus-4-6|gemini-3.1-pro|gpt-5.2-deep|gpt-5.2-chat|human,
  provenance: {channel: web_search|file_search|site_restricted|quick_validation|internal_document,
    site_restrictions: [], context_documents: [], citation_quality: 0-5, temporal_marker: "YYYY-MM-DD"|null},
  depends_on: [claim IDs]}
```

### Citation Quality Scale

| Score | Source Type | Examples |
|-------|------------|---------|
| 5 | Primary | SEC filings, peer-reviewed, official stats, vendor docs |
| 4 | High-quality secondary | Gartner, Forrester, named-source journalism |
| 3 | General secondary | News, press releases, industry publications |
| 2 | Tertiary | Wikipedia, blogs, aggregators |
| 1 | Unsourced assertion | No citation trail |
| 0 | Unverifiable | Contradicts known facts or cites non-existent sources |

---

## Step 3: Triage Claims

Decompose normalized claims into an atomic claims matrix with dependency tracking.

### Claim Types

| Type | Handling |
|------|----------|
| **Factual assertions** | Cross-validate; trace to primary sources |
| **Causal claims** | Map reasoning chains; note mechanism divergence |
| **Quantitative data** | Flag discrepancies >10%; verify primary source agreement |
| **Recommendations** | Tag as interpretation; link to supporting facts |
| **Unique insights** | Preserve with provenance flag; do not discard |
| **Reasoning chains** | Preserve structure; validate logical steps; compare across sources |
| **Structural artifacts** | Preserve format (tables, matrices, timelines); merge in Step 5 |

### Claims Matrix

Rows = claims, columns = sources. `(qN)` = citation quality. Flag same-primary-source convergence separately.

```
             │ Claude │ Gemini │ GPT-5.2D│ GPT-5.2C│ Status
─────────────┼────────┼────────┼─────────┼─────────┤
Claim A      │ ✓ (q4) │ ✓ (q5) │  ✓ (q3) │         │ Convergent (diff sources)
Claim B      │ ✓ (q4) │ ✓ (q4) │         │         │ Convergent (same primary?)
Claim C      │ ✓ (q3) │        │         │         │ Unique
Claim D      │ ✓ (q4) │ differ │  differ │         │ Contested
```

### Claim Dependency Graphs

`Claim A (factual) ──supports──► Claim B (causal) ──supports──► Claim C (recommendation)`

**Rules**: Recommendation confidence CANNOT exceed supporting facts. If a fact drops tiers, propagate to ALL dependents. Document chains in output.

---

## Step 4: Reconcile

### Provenance-Weighted Disagreement Protocol

Execute in order — stop at first resolution:

| Step | Test | Resolution |
|------|------|-----------|
| 1 | **Coexist?** Different scope/timeframe/definition? | Preserve both with clarifying context |
| 2 | **Provenance?** Higher channel? (`site_restricted` > `web_search` > `quick_validation`; `internal_document` > `web_search` for proprietary) | Favor higher provenance; note alternative |
| 3 | **Citation dedup?** Same primary source(s)? | Treat as SINGLE-SOURCE (Tier 2 cap), not independent |
| 4 | **Specificity?** One more specific or better sourced? | Favor specificity; note alternative |
| 5 | **Majority?** Most sources agree? | Lead with majority; preserve dissent |
| 6 | **All diverge** | Flag "unresolved." Agentic mode: web search. Else: present all views |

### Confidence Tier