VividKit Crystal clear AI coding
VI
Choose Your Setup Get Started with AgentKit Continue with ClaudeKit Migrate from ClaudeKit to AgentKit Clean cutover from ClaudeKit What Is AgentKit? How AgentKit Works AgentKit Kits, Skills, Agents & Hooks AgentKit Runtime Support AgentKit Desktop App AgentKit Configuration AgentKit CLI Quick Reference AgentKit Skills Cheatsheet AgentKit Workflows AgentKit Statusline Update AgentKit CLI & Kits AgentKit Helper Clean reinstall of AgentKit AgentKit Troubleshooting What is ClaudeKit? ClaudeKit CLI Guide Command Finder Commands Cheat Sheet Workflow Recipes CLI Quick Reference Claude Mechanics Permission Modes ClaudeKit Hooks Coexistence ClaudeKit x Codex CK → Other coding tools How ClaudeKit Works Inside ClaudeKit Finding Your Unknowns Session Recovery Fix from Logs UI/UX Guide Accounts Switcher IDE Config Remote Control Codex App System Diagramming UI Review Gate Promotions Support VividKit

Quick Ref / Command

Engineer Kit

/ak:research

Source-backed technical research

Run concise, source-backed technical research from scope definition through multi-source gathering, synthesis, recommendations, and saved report.

01

Define scope

02

Plan searches

03

Gather evidence

04

Analyze deeply

Rule 01

KISS and DRY guide scope and recommendations

Rule 02

Current, authoritative sources beat volume

Rule 03

Security and performance research must check recent advisories and benchmarks

Rule 04

--ultra is a best-of-5 verifier pass, not a different research algorithm

Execution Map

Core lanes from input to output, aligned with the command's real execution path.

01

Start

Input, scope, route

  1. 1 Define scope Name key concepts, recency requirements, source evaluation criteria, and research boundaries.
  2. 2 Plan searches Craft precise queries for official docs, repositories, authoritative blogs, changelogs, security, and performance evidence.
02

Work

Agent / skill execution

  1. 3 Gather evidence Use native web_search for current external research, running independent queries in parallel when allowed and never exceeding five calls.
  2. 4 Analyze deeply Read official docs, API references, specs, README files, releases, and video sources when relevant.
03

Verify

Gate, review, validation

  1. 5 Cross-check claims Validate across independent sources, note dates, consensus, conflicts, deprecations, and migration paths.
  2. 6 Synthesize report Produce an executive summary, methodology, findings, comparative analysis, recommendations, resources, and appendices as needed.
04

Close

Report, handoff, artifact

  1. 7 Organize output Invoke project-organization to organize outputs, then save the report at the Naming section's Report path; ask the main agent if that path is missing.

Syntax, arguments, subcommands

How to invoke the skill: syntax, positional arguments, shared options, then each subcommand with its own syntax and outcome.

Syntax

/ak:research [topic] [--ultra] [--yagni]

Arguments

[topic] Required

Research question

Free-form technical decision or comparison to research. Include alternatives, constraints, criteria, recency window, research-call budget, and the approved Report path when available; this argument does not authorize implementation or publication.

/ak:research "Compare Postgres RLS and service-layer authorization for a multi-tenant B2B API; max 4 research calls; Report: plans/research/tenant-authorization.md"

Shared options

--ultra

Best-of-5 verifier

Runs five independent read-only candidate researchers from one evidence packet, then a verifier selects one report unchanged or rejects all. It keeps the per-run research-call budget.

/ak:research OAuth device flow security --ultra
--yagni

Cut unneeded scope

Challenges and removes research scope that is not needed for the stated decision or outcome. It must not drop anything the user explicitly requires.

/ak:research vector database options for 10k docs --yagni

Modes / Routes

default

Single research report

Research:
One controlled multi-source pass
Review:
None
Validate:
Cross-source verification
--ultra

Best-of-5 verifier

Research:
Five independent candidate reports from one evidence packet
Review:
None
Validate:
Strongest verifier ranks and selects or rejects
--yagni

Scope cutter

Research:
Challenges unneeded scope
Review:
None
Validate:
Only keeps work needed for stated outcome

Sample Prompt

Concrete invocations for each flag, subcommand, mode, or route available in the live workflow.

Technical evaluation Recommended 
/ak:research choose a durable queue for Next.js background jobs
Use when:
Use before implementation when architecture, compatibility, security, and maintenance tradeoffs matter.
Expected:
A markdown report that defines scope, cites authoritative sources, compares queue options, calls out security/performance tradeoffs, and ends with concrete next steps.
Verifier research 
/ak:research OAuth device flow security --ultra
Use when:
Use when the answer is high-risk and worth five independent candidate passes.
Expected:
Five independent candidate reports are scored by a verifier, which either preserves the winning report with a ranking appendix or stops with reject-all reasoning.
YAGNI research 
/ak:research vector database options for 10k docs --yagni
Use when:
Use when the likely risk is over-building beyond the stated outcome.
Expected:
A focused report that challenges excess scope, keeps only research needed for the stated decision, cites sources, and lists unresolved questions.

Handled Scope

  • Define scope
  • Plan searches
  • Gather evidence
  • Analyze deeply
  • Cross-check claims
  • Synthesize report

Artifacts Produced

Research report

Pattern

Report path from the Naming section; if absent, ask the main agent for the output path

Markdown report with sources, methodology, findings, recommendations, references, and unresolved questions.

Next