researchkit

Compare technical options against primary sources and recommend one with cited evidence.

1|Updated Jul 12, 2026
One-click install
npx skills add https://github.com/mimukit/skills --skill researchkit-mimukit
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: researchkit
Source: https://github.com/mimukit/skills/tree/main/skills/researchkit
Command: npx skills add https://github.com/mimukit/skills --skill researchkit-mimukit

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Choosing between tools, libraries, frameworks, or architectures often means wading through stale blog posts and opinion pieces. This Skill researches the credible options for a technical decision using primary sources and lands a single recommendation with cited, dated evidence. ## Core Features & Use Cases - Decision framing: Pins down the constraints that actually decide the choice (stack, scale, budget, team familiarity) before comparing anything. - Primary-source investigation: Reads official docs, source code, specs, and maintainer benchmarks, recording version and date for every load-bearing claim and flagging stale evidence. - Recommendation with handoff: Produces a structured artifact (recommendation, comparison table, evidence list, open questions) and optionally saves it to docs/research/ or hands off to a planning skill. - Use Case: Ask "Drizzle or Prisma for this project?" and receive a cited comparison on the constraints that matter, a clear pick, and the condition under which you'd choose differently. ## Quick Start Ask the agent to research which queue library to use for your workload and recommend one with cited sources.

Frequently Asked Questions about researchkit

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I compare two libraries before choosing one?

Frame the decision constraints first (stack, scale, budget, must-have features), then investigate each option against primary sources like official docs and source code. The output is a comparison table on those constraints plus a single recommendation with cited, dated evidence.

What sources does technical decision research use?

It relies on primary sources: official documentation, source code, specifications, first-party APIs, and maintainer-published benchmarks. Each load-bearing claim carries its source URL with version and date, and stale evidence is explicitly flagged.

Can it build a prototype or spike to test a hypothesis?

No. It never writes, runs, or scaffolds code to settle a claim. Unresolvable questions are marked as open questions for a separate, explicitly requested prototyping step, keeping the deliverable a cited argument rather than code.

What happens when there is no web access available?

It states the limitation plainly and gives a best-effort comparison from existing knowledge with an explicit staleness warning. It never fabricates citations; a missing source is reported as missing.

Where is the research report saved?

By default the report prints inline only. On request it is written to docs/research/research-<slug>-YYYY-MM-DD.md, or to whatever research or RFC location and naming scheme the repository already uses.