How does the skill handle conflicting information between sources?

The skill is designed to surface conflicts explicitly rather than choosing one version. It presents the differing viewpoints along with their respective sources and timestamps so the user can make an informed judgment.

How does the confidence scoring work?

Confidence is calculated by balancing 'Freshness' (how recently the information was updated) against 'Authority' (the reliability of the source, such as official docs vs. informal chat).

What types of sources can this skill synthesize?

It is optimized to handle a wide variety of enterprise data including chat messages, emails, cloud storage documents, project tracker tasks, and official wiki entries.

Will I still be able to see the original source of the information?

Yes. Every claim in the synthesized answer includes inline citations and a comprehensive source list at the end, detailing titles, authors, and specific locations.

Knowledge Synthesis for Enterprise Search

Name: Knowledge Synthesis for Enterprise Search
Author: anthropics

byanthropics

•

8,000

•

Productivity & Workflow

Transforms fragmented search results from multiple enterprise sources into coherent, cited answers with automated deduplication and confidence scoring.

This skill acts as the critical final step in enterprise search by aggregating raw data from disparate sources—such as chat threads, emails, and cloud storage—into a single, trustworthy narrative. It intelligently identifies and merges duplicate information while maintaining clear provenance through structured citations. By evaluating source authority and data freshness, the skill assigns confidence levels to its findings, ensuring that users can distinguish between finalized project decisions and informal, potentially outdated discussions.

Key Features

01Automated source attribution including author, date, and specific location (e.g., #channel or folder)

02Conflict detection that explicitly flags and summarizes contradictory information across sources

038,000 GitHub stars

04Adaptive summarization strategies for small, medium, and large result sets

05Cross-source deduplication and information merging based on temporal and contextual signals

06Dynamic confidence scoring weighted by source authority and information recency

Use Cases

01Reconstructing the timeline and final resolution of project decisions from fragmented chat and email history

02Generating 'source of truth' summaries from vast internal document repositories and meeting notes

03Onboarding new team members by synthesizing current project status from diverse knowledge silos

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add anthropics/knowledge-work-plugins knowledge-synthesis

For use in Claude.ai and ChatGPT

Download Skill