summarization

Condense prose into faithful summaries that preserve evidence and decisions.

1|Updated May 6, 2026
One-click install
npx skills add https://github.com/jacob-balslev/skill-graph --skill summarization-jacob-balslev
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: summarization
Source: https://github.com/jacob-balslev/skill-graph/tree/main/marketplace/skills/summarization
Command: npx skills add https://github.com/jacob-balslev/skill-graph --skill summarization-jacob-balslev

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Prevents long agent outputs and research artifacts from becoming technically complete but practically unreadable by turning them into meaning-preserving summaries.

Core Features & Use Cases

  • Faithful semantic condensation that preserves key claims, decisions, evidence links, and uncertainty.
  • Summary-level control (TLDR, executive summary, brief, condensed report) to match audience and constraint.
  • Progressive summarization and handoff readiness so successors can continue work without reopening the full source.

Quick Start

Ask the agent to write an L2 executive summary of the audit findings for the workspace scope while preserving all evidence links and the full finding counts.

Frequently Asked Questions about summarization

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

FAQPage Schema
How do I write an executive summary that preserves evidence links and uncertainty?

An executive summary requires faithful semantic condensation that preserves key claims, decisions, evidence links, and uncertainty. You must select an explicit summary level and enforce a faithfulness gate to prevent invented claims or flattened uncertainty.

What is progressive summarization for agent handoffs?

Progressive summarization for agent handoffs condenses long research artifacts into meaning-preserving briefs. It ensures successor agents can continue work without reopening the full source by retaining evidence paths and actionable next steps.

How do I condense audit reports without losing provenance and finding counts?

Audit report condensation applies a faithfulness gate to lossy reduction, keeping the process auditable. It requires selecting an explicit summary level while retaining evidence paths, full finding counts, and provenance for compliance.

Can I generate a TLDR from long agent outputs while keeping decisions and next steps?

Generating a TLDR from long agent outputs uses summary-level control to match audience constraints. It preserves key claims, decisions, and actionable next steps, preventing technically complete outputs from becoming practically unreadable.

What's the best way to summarize research notes for agent workflows?

Research-to-notes distillation uses semantic condensation to turn long research artifacts into faithful summaries. It enforces a faithfulness gate to ensure no invented claims and retains evidence paths for agent workflows.

When should I not use lossy reduction for session wrap reports?

Lossy reduction for session wrap reports should not be used when auditable traceability is required. If you cannot enforce a faithfulness gate or retain evidence paths and provenance, the summary may flatten critical uncertainty or lose finding counts.