knowledge-synthesis_skill

This skill synthesizes multi-source results into coherent, deduplicated answers with source attribution, delivering concise, trustworthy summaries for informed
  • Python
  • Official

7.4k

GitHub Stars

1

Bundled Files

2 months ago

Catalog Refreshed

4 months ago

First Indexed

Readme & install

Copy the install command, review bundled files from the catalogue, and read any extended description pulled from the listing source.

Installation

Preview and clipboard use veilstrat where the catalogue uses aiagentskills.

npx veilstrat add skill anthropics/knowledge-work-plugins --skill knowledge-synthesis

  • SKILL.md8.3 KB

Overview

This skill combines search results from multiple sources into a single, coherent answer with deduplication, source attribution, and confidence scoring. It produces concise narratives that merge duplicate facts, surface conflicts, and preserve important nuance. The output is tuned for knowledge workers who need trustworthy, actionable summaries across chats, email, docs, and trackers.

How this skill works

The skill ingests raw results from varied sources, deduplicates overlapping content using signals like identical text, author, timestamps, and explicit cross-references, and chooses the most complete, authoritative, or recent version when merging. It clusters related items, ranks clusters by relevance, assesses confidence using freshness and source authority, then synthesizes a narrative answer with inline attributions and a source list. For large result sets it provides a high-level summary with counts and offers drill-down options.

When to use it

  • When you need a single trustworthy summary from chat threads, emails, docs, and tickets
  • When multiple sources repeat the same fact and you want a deduplicated narrative
  • When you must see source attributions and confidence for each claim
  • When search returns many results and you need grouped themes and top sources
  • When sources conflict and you want the disagreement surfaced rather than hidden

Best practices

  • Prefer the most complete, authoritative, or recent source when merging duplicate facts
  • Always include source type, location, date, and author where relevant
  • Flag low-confidence claims and avoid presenting outdated info as current
  • Group related findings by theme rather than listing results source-by-source
  • Offer drill-down detail for large result sets instead of dumping raw items

Example use cases

  • Summarize a product decision that appears across Slack, email, and the spec doc with confirmation and timeline
  • Produce a status snapshot for a project by merging ticket updates, meeting notes, and shared documents
  • Detect and surface conflicting guidance (e.g., policy versions) and show which source is most recent
  • Aggregate customer feedback from chat transcripts, support tickets, and survey results into top themes
  • Turn a large set of search hits into a concise executive summary with top supporting sources

FAQ

Confidence combines freshness (recency), source authority (official docs > email > chat), and agreement across sources. The skill reports high, moderate, or low confidence and explains the factors.

Will duplicate information be removed?

Yes. The skill merges duplicate facts into a single narrative item, cites all sources where it appeared, and preserves unique details from each source.

Built by
VeilStrat
AI signals for GTM teams
© 2026 VeilStrat. All rights reserved.All systems operational
knowledge-synthesis skill by anthropics/knowledge-work-plugins | VeilStrat