Orchestrate web research into claim-backed reports with verifiable citations.
npx skills add https://github.com/thistleknot/skills --skill deep-research-thistleknot
Please help me install this Agent Skill. Skill: deep-research Source: https://github.com/thistleknot/skills/tree/main/deep-research Command: npx skills add https://github.com/thistleknot/skills --skill deep-research-thistleknot
Deep-research provides a repeatable protocol for gathering multi-source web evidence and turning it into claim-backed, citation-traceable reports when prior memory or lightweight retrieval is insufficient.
(subject, predicate, object) triplets tied to minted source_ids, preserving a verifiable citation chain.Use deep-research to produce a markdown research report with citations by running it on your research question and allowing it to fetch and extract evidence across multiple independent domains.
High-intent search queries and answers about installing and using this skill.
To gather web evidence and compile a research report with verifiable citations, you need a workflow that extracts structured triplets tied to minted source IDs. This orchestrates parallel subquestion decomposition and cross-source corroboration to ensure claims are traceable.
Evidence saturation detection identifies diminishing returns during multi-source web research by applying saturation thresholds. It enforces minimum domain and source coverage before final synthesis, ensuring comprehensive corroboration.
Triplet extraction preserves a citation chain by storing subject-predicate-object relationships tied to minted source IDs. This creates a verifiable citation graph linking every extracted claim back to its original web evidence source.
Yes, agent orchestration for deep research requires a tiered MCP tool suite. You need search, fetch, and vector retrieve or upsert capabilities to execute the L0 web evidence cascade and maintain enforced source_id integrity.
Use a retrieval pipeline with parallel subquestion decomposition when prior memory or lightweight retrieval is insufficient for your research task. It applies to multi-source corroboration scenarios requiring structured, claim-backed reports with verifiable citations.
The best way to structure multi-source corroboration for a citation graph is extracting structured triplets tied to source IDs. This dispatches concurrent researcher runs, reflects on coverage, and synthesizes findings into a final markdown report.
haxlysPerform top-k similarity search over vector embeddings across Pinecone, Qdrant, pgvector, and Weaviate.
JonasTischerManage academic research papers from discovery to citation verification.
leafcoderIntegrate retrieval-augmented generation and agentic orchestration frameworks for LLM applications.
vladkeslerFetch web pages and HTTP APIs to produce cited summaries.