Skip to content
Back to skills

Synthesizing Research

ASecurity

Synthesize all prior research into a unified GTM narrative. Run after `analyzing-contradictions` to prepare for slide deck generation (Phase 5).

  • 2 stars
  • 0 votes
  • 0 copies
  • 2 views
  • Added September 5, 2026
ai-agents

Works with

  • claude code

Security analysis

A100/100

Scanned September 29, 2026

npx -y skills add qte77/claude-code-plugins --skill synthesizing-research --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Synthesizing Research?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Synthesizing Research
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/qte77-synthesizing-research/badge)](https://www.skillsdirectory.com/skills/qte77-synthesizing-research)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: synthesizing-research
description: Synthesize all prior research into a unified GTM narrative. Run after `analyzing-contradictions` to prepare for slide deck generation (Phase 5).
compatibility: Designed for Claude Code
metadata:
  stability: stable
  content-hash: sha256:a8c1dc87f705a039a21f114e8b7f28a7f8518948388e7ed17f573ff5949b2160
  allowed-tools: Read, Write, Glob, Grep, WebSearch, WebFetch
  argument-hint: [optional-synthesis-focus]
---

# Synthesizing Research (Phase 5)

**Target**: $ARGUMENTS

Phase 5 of the GTM pipeline. Depends on all prior phases (0-4). Produces a
unified GTM narrative integrating technical, market, and strategic findings.
This output drives Phase 6 slide deck generation.

## Mode Awareness

Read `config/mode.md` before starting:

- `concise` — Executive summary (1 page), key findings table, single recommended path
- `detailed` — Full integrated narrative, alternative scenarios, confidence annotations
- `conservative` — Lead with validated evidence; clearly separate confirmed from inferred
- `ambitious` — Lead with opportunity thesis; build the bull case narrative

## Inputs

- `results/phase-0/capability-profile.md`
- `results/phase-1a/competitor-map.md`
- `results/phase-1b/market-analysis.md`
- `results/phase-2/pmf-assessment.md`
- `results/phase-3/gtm-strategy.md`
- `results/phase-4/contradiction-analysis.md`
- `config/comments_gtm.md` — GTM preferences for framing
- `config/mode.md` — Style and approach settings

## Workflow

1. **Read all prior phase outputs** including contradiction resolutions from Phase 4
2. **Construct the narrative arc**: Source capability → Market opportunity → PMF evidence → GTM path
3. **Resolve outstanding contradictions** from Phase 4 into final recommendations
4. **Assign confidence ratings** to all major claims (High/Medium/Low)
5. **Identify the single recommended GTM path** and articulate why
6. **Document alternatives** with tradeoffs for decision-maker context
7. **Validate against criteria** — Check `config/validation_criteria.md` Phase 5 gates

## Output

Write to `results/phase-5/`:

### `synthesis.md`

```markdown
# GTM Research Synthesis

## Executive Summary
[3-5 sentence narrative: what we found, what it means, what we recommend]

## Key Findings by Phase

| Phase | Key Finding | Confidence |
|-------|-------------|------------|
| 0 — Source | [finding] | High/Medium/Low |
| 1A — Landscape | [finding] | High/Medium/Low |
| 1B — Market | [finding] | High/Medium/Low |
| 2 — PMF | [finding] | High/Medium/Low |
| 3 — GTM | [finding] | High/Medium/Low |
| 4 — Gaps | [finding] | High/Medium/Low |

## Integrated GTM Narrative

### The Opportunity
[Market context and timing rationale]

### The Fit
[How product capabilities match market needs]

### The Path
[Recommended GTM motion and why]

## Recommended GTM Path
[Specific, actionable recommendation with first 3 steps]

## Alternative Paths
### Alternative 1: [Name]
- Tradeoff: [upside vs downside]

## Confidence Assessment
- Claims with High confidence: [N]
- Claims with Medium confidence: [N]
- Claims with Low confidence: [N] — [action needed]
```

## Rules

- Low-confidence claims must be explicitly flagged — do not bury uncertainty
- Contradiction resolutions from Phase 4 must be reflected in the narrative
- The recommended path must be specific enough to act on within 30 days
- Reference phase output files by path for traceability

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.