Research Routing and Cited Briefings
Route research to the right method, then produce a cited briefing with evidence gaps, disagreements, and a source audit.
Before & after
Same prompt, same agent — the only difference is the skill.
I can help you research AI agent frameworks. That's a broad topic, so let me gather information from multiple sources and synthesize what's out there. I'll look at public documentation, GitHub repositories, academic papers if relevant, and maybe some recent articles. There's a lot of activity in this space right now, so the landscape is shifting pretty quickly. I can put together a summary of the main frameworks, their strengths and weaknesses, and cite sources so you can dig deeper. What specific aspects matter most — performance, ease of use, open-source vs commercial, scalability?
AI AGENT FRAMEWORKS — RESEARCH BRIEFING MARKET OVERVIEW: 15+ active frameworks launched in 2024-2025. Landscape fragments into orchestration platforms (Claude Agents, LangChain agents, Anthropic SDK) vs lightweight libraries (pydantic-ai, controlflow). No clear market leader yet; adoption driven by existing ML ecosystem inertia. TOP 5 FRAMEWORKS: 1. Claude Agents (Anthropic) — GitHub stars: private API, adoption via claude.ai/SDK Use case: multi-step reasoning, structured tool use Trade-off: vendor lock-in, pricing per tool call 2. LangChain Agents — Stars: 70k+ Use case: LLM app scaffolding, multi-provider Trade-off: opinionated abstractions, learning curve 3. Anthropic SDK (client.beta.messages) — integrated agents, low-level control 4. Pydantic-AI — lightweight validation-first agent pattern 5. Autogen (Microsoft) — multi-agent conversation framework USE CASE MAPPING: • Long-horizon reasoning: Claude Agents • Multi-LLM fallback: LangChain • Production inference cost optimization: bare SDK TRADE-OFF MATRIX: latency (Claude Agents slowest), cost (per-token vs per-call), lock-in, observability. CITATIONS: docs.anthropic.com, langchain.com, GitHub stars snapshot 2026-07-31.
About this skill
“Research this” can produce a pile of links without answering the question. This workflow first identifies the kind of research needed, then builds a cited briefing that separates established findings, competing explanations, and unanswered questions.
What it does
- Classifies requests into trends, funding, academic literature, course readings, patent research, entity research, or general research.
- Explains the selected route and proceeds even when no specialist workflow is installed.
- Breaks broad questions into answerable subquestions with explicit scope and stopping conditions.
- Checks source independence, publication dates, evidence quality, and whether citations actually support each claim.
- Investigates disagreements instead of silently selecting the most convenient source.
- Produces a concise synthesis with traceable citations and a compact search audit.
What you get
A Markdown research briefing sized to the request, with an answer up front, supporting evidence, uncertainties, source links, and research limitations. When live search is unavailable, you get a clearly labeled provisional assessment or research plan instead of invented findings.
Best for
Analysts, founders, writers, and teams handling varied research requests who need the right research method before committing to a full investigation.
SKILL.md preview
---
name: research-routing-and-cited-briefings
description: Classifies research requests by their intended deliverable, selects an appropriate specialist method, and produces traceable multi-source briefings using claim-level evidence checks. Use when the user says "research this topic", "look into this for me", "what do we actually know about this?", "investigate these claims", "I keep getting conflicting answers", "turn these sources into a briefing", or "I have a pile of links but no answer".
---
# Research Routing and Cited Briefings
… (sign up to view the full skill)More research & data skills
View all Research & Data skills →Source-Reconciled Research Brief
Turn supplied links into a concise brief with claim-level citations, reconciled source conflicts, and explicit evidence gaps.
Source-Traced Research Bundle
Compile many sources into one bundle where every claim is atomic, sourced, dated, tiered, and conflict-checked, in Markdown or JSON.
A/B Test Power and Validity Designer
Exact A/B sample sizes and MDE with math shown, plus the traps: unit mismatch, multiple variants, peeking, guardrail power, and SRM.
X/Twitter Research with Cited Post Evidence
Research X topics, threads, and accounts with cited posts, deliberate sampling, and clear limits on sentiment claims.