synthesize

Restructure agent output into scannable, purpose-matched formats using need-based routing templates.

14|1|Updated Jun 14, 2026
One-click install
npx skills add https://github.com/allemaar/open-skills --skill synthesize-allemaar
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: synthesize
Source: https://github.com/allemaar/open-skills/tree/main/skills/synthesize
Command: npx skills add https://github.com/allemaar/open-skills --skill synthesize-allemaar

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve? Agent output often arrives as dense walls of text that bury decisions, caveats, and required actions. This Skill restructures that output into a reader-need-matched shape without changing its meaning, so every protected fact, number, and uncertainty survives the rewrite. ## Core Features & Use Cases - Need-based routing: Routes output to one of 23 reader-need templates (answer, decide, compare, status, timeline, handover, and more) and loads only the selected reference file. - Fidelity contract: Inventories protected units (negations, numbers with units, attributions, deadlines, reversing conditions) before composing and labels every claim as confirmed, judgement, or estimate. - Render grammar: Applies a shared grammar of status labels, cues, tables, and timelines, with optional handoff to a figure-drawing skill when position or proportion matters. - Use Case: After a long investigation report, ask the agent to synthesize it for a decision; the output leads with the bottom line, keeps both sides of any disagreement attributed, and closes with the next action and its deadline. ## Quick Start Use the synthesize skill to restructure this investigation report into a decision-ready summary for my manager.

Frequently Asked Questions about synthesize

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I make long AI agent output easier to read?

Route the output through a need-based synthesis pass that identifies what the reader must do, then restructures the text into a matching shape such as a decision brief, comparison table, or status report. The process preserves every material fact, number, and caveat while cutting repetition.

How to summarize AI output without losing important details?

Inventory the protected units first: negations, numbers with units, attributions, deadlines, and conditions that could reverse the conclusion. Then restructure around those units so none appear only below the fold, and label each claim as confirmed, judgement, or estimate.

What output formats does the synthesize skill support?

It supports 23 reader-need shapes including answer, point, learn, investigate, advice, decide, compare, prioritize, act-now, procedure, plan, status, monitor, change, timeline, organize, review, feedback, rewrite, explore, handover, record, and a hybrid mode for multiple needs.

Can synthesize handle conflicting information from multiple sources?

Yes, but it never merges conflicting sources silently. Each source stays attributed with its stance, and disagreements render with both sides visible. When several reports must be reconciled first, it defers to a dedicated merge skill if installed.

When should I not use output synthesis?

Skip it when a simple answer is already minimal and clear, since a no-op is a valid result. It also does not verify facts, so restructuring never raises confidence above the source material, and deep writing craft belongs to separate writing skills.