3
GitHub Stars
2
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 krishagel/geoffrey --skill research- RESEARCH_ENHANCEMENT.md4.9 KB
- SKILL.md11.1 KB
Overview
This skill delivers discovery-driven, exhaustive research by running multiple LLM researcher agents in parallel and synthesizing their findings. It focuses on relevance to your context, deep source discovery (including multimedia and forums), and strict citation for every factual claim. Reports are saved to a configured document location for review and follow-up.
How this skill works
The skill decomposes complex queries into 3–7 targeted sub-queries (factual, comparative, current state, user experience, edge cases) and launches parallel agents tuned for different strengths (current web retrieval, synthesis, structured analysis, deep reasoning). It gathers web, multimedia, and forum content, escalates to a browser-driven fetch when pages need JS rendering or authenticated access, and produces a synthesized report with inline citations and a complete source list. Confidence scores, conflicting evidence, and actionable recommendations are included.
When to use it
- You need current, verifiable information (deals, policies, perks).
- Comparing options across multiple perspectives or vendors.
- Deep investigation of complex or rapidly changing topics.
- Fact-checking, verification, or source-backed claims.
- Market research, trend analysis, or long-form briefs.
Best practices
- Start by answering clarifying questions so research targets real goals and constraints.
- Provide any relevant preferences or account context so domain loading customizes results.
- Allow longer run times for exhaustive work — 30 minutes to several hours for deep dives.
- Require browser-driven fetching for forum threads, authenticated pages, or dynamically rendered content.
- Expect every factual claim to include a citation; mark items as unverified if no source is found.
Example use cases
- Maximizing airline miles to a specific destination with partner and award-availability patterns and real traveler reports.
- Comparing business tools with current pricing, user reviews, expert analyses, and multimedia demos.
- Investigating product safety or regulatory changes with primary sources and expert commentary.
- Market landscaping for a niche startup idea: trends, competitor signals, and first-hand user feedback.
- Travel planning that includes user reviews, trip reports, transport tips, and "what I wish I knew" sections.
FAQ
Shallow lookups can be minutes, but exhaustive reports typically take 30 minutes to 3 hours depending on depth and the need for browser-controlled crawling.
Will every claim include a source?
Yes. Every factual claim requires an inline citation. If a claim cannot be sourced it is flagged as unverified or omitted until a source is found.