Synero Skill

Use this skill to query Synero from the terminal when one AI answer is not enough. Synero is positioned as a collective AI intelligence product: one prompt goes to four advisors with different reasoning styles, then Synero synthesizes them into a single response.

This is most useful for ambiguous or high-leverage questions where you want disagreement, tradeoffs, blind-spot checking, and a clearer final recommendation.

What this skill helps with

Use it for prompts like:

• - product or roadmap decisions with meaningful tradeoffs
• technical architecture reviews and go/no-go questions
• research questions where cross-checking perspectives improves trust
• hiring, org, or leadership choices with second-order effects
• messaging, positioning, or content strategy that benefits from competing angles

Prefer simpler tools for basic factual lookups or one-shot chores. This skill shines when the question is judgment-heavy.

What it does

• - Sends a prompt to Synero’s council endpoint
• Supports 4 advisor roles: architect, philosopher, explorer, INLINECODE3
• Returns either:

• - Supports optional thread continuity and per-slot model overrides

Prerequisites

Get your API key from https://synero.ai, then export it before running the script:

CODEBLOCK0

If you are not sure where to get the key, sign in at https://synero.ai and use the API/settings area there.

Optional environment variables:

CODEBLOCK1

Quick command

CODEBLOCK2

That command uses SYNERO_API_KEY from your environment and sends the request to Synero at https://synero.ai/api/query unless you override SYNERO_API_URL.

Quiet final-output mode

Use --quiet when you want only the synthesized answer with no extra status lines:

CODEBLOCK3

Streaming and debugging mode

Use --raw when you want the raw event stream for troubleshooting or to inspect council behavior live:

CODEBLOCK4

Advanced configuration

Use model overrides when you want to pin specific advisors or a specific synthesizer:

CODEBLOCK5

Output behavior

Default mode prints:

• - HTTP status line
• final synthesized answer

INLINECODE11 prints:

• - final synthesized answer only

INLINECODE12 prints:

• - raw SSE events from the API

Prompting guidance

Write prompts that invite useful disagreement, not generic brainstorming.

• - Ask for tradeoffs, risks, assumptions, and a recommendation
• Include operating constraints such as timeline, budget, team size, and risk tolerance
• Use --thread-id when continuing the same topic across multiple rounds
• Use --quiet when another tool or script needs clean final text only
• Use --raw when debugging streaming behavior or inspecting advisor outputs

For reusable prompt templates, read:

• - INLINECODE16

Error handling

• - Missing key → exits with clear guidance to set INLINECODE17
• HTTP failure → prints status and response body
• Network failure → prints a clear network error
• Empty synthesis → exits non-zero instead of pretending things are fine