Research Skills
The synth-* family gathers, compares, and synthesises information from multiple sources or model perspectives. Output feeds strategic and architectural decisions.
Skills
Section titled “Skills”| Skill ID | Description | Model Class |
|---|---|---|
synth-comparative | Structured comparison of N options across defined evaluation dimensions | free |
synth-research | Deep-dive research on a topic: collects evidence, identifies consensus and controversy | free |
synth-engine | Research synthesis engine: takes multiple raw research inputs and produces a unified report | strong |
synth-recommendation | Generates a decision recommendation with supporting rationale from research outputs | cheap |
Pattern 5 — Free Triple Parallel
Section titled “Pattern 5 — Free Triple Parallel”Research Request ├── GPT-5 mini → perspective A (broad) ├── GPT-4.1 → perspective B (thorough) └── GPT-4.1 → perspective C (alt framing) │ └── synth-engine (Claude Sonnet 4.6) → unified synthesis │ └── synth-recommendation → decision outputCost: 3 free calls + 1 synthesis pass ≈ 80% cheaper than strong end-to-end.
When to Use
Section titled “When to Use”| Situation | Skill(s) |
|---|---|
| Choosing between frameworks/tools | synth-comparative + synth-recommendation |
| Understanding an unfamiliar domain | synth-research + synth-engine |
| Building a justified recommendation | All four |
Instructions That Invoke These Skills
Section titled “Instructions That Invoke These Skills”- research — primary consumer; all four coordinated with Pattern 5
- strategy — uses
synth-comparative+synth-recommendationfor tradeoff analysis - design — uses
synth-researchfor technology selection
synth-comparative Format
Section titled “synth-comparative Format”## Comparison: Option A vs Option B vs Option C
| Dimension | Option A | Option B | Option C ||-----------|----------|----------|----------|| Performance | ★★★★☆ | ★★★☆☆ | ★★★★★ || ...
**Recommendation**: Option C for high-throughput, Option A for ease of maintenance.