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:docs-seeker

Find current library docs

Script-first discovery for current library, framework, API, and repository documentation through llms.txt and context7.

01

Read Query

02

Detect Topic

03

Fetch Docs

04

Analyze Coverage

Rule 01

Scripts first, manual URL construction last.

Rule 02

Zero-token scripts do classification, fetching, fallback, and distribution planning.

Rule 03

Progressively load workflow references only when the query needs them.

Rule 04

Current docs beat memory for APIs, features, and repository behavior.

Execution Map

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

01

Start

Input, scope, route

  1. 1 Read Query Capture the library or framework and the topic the user needs, such as an API, feature, or repository behavior.
  2. 2 Detect Topic Run detect-topic.js to classify topic-specific versus general documentation and extract library plus topic.
02

Work

Agent / skill execution

  1. 3 Fetch Docs Run fetch-docs.js with the original query so scripts handle context7 URL construction, fallback chains, and errors.
  2. 4 Analyze Coverage When fetched llms.txt content contains multiple URLs, pipe the content to analyze-llms-txt.js for critical/important/supplementary categories and agent distribution.
03

Verify

Gate, review, validation

  1. 5 Load Targeted References Open topic-search, library-search, repo-analysis, context7-patterns, errors, or advanced guidance only when needed.
  2. 6 Read Primary Pages Use returned URLs and web reading to ground answers in current documentation rather than memory.
04

Close

Report, handoff, artifact

  1. 7 Distribute if Broad Follow the script's recommended strategy: one agent, three agents, seven agents, or phased research.
  2. 8 Answer with Sources Summarize the relevant API or feature with links, errors encountered, fallback path, and any version or edge-case notes.

Syntax, arguments, subcommands

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

Syntax

/ak:docs-seeker [library-name] [topic]

Arguments

[library-name] Required

Library or framework

Name the package, framework, API, docs site, or repository whose current documentation should be found.

/ak:docs-seeker "Next.js 16 cache invalidation APIs"
[topic]

Topic

Optional feature, API, version, target language, or question to narrow the search. Omit it for broader library coverage that may need URL prioritization.

/ak:docs-seeker "shadcn date picker"

Special Operations

Topic search

For queries like a date picker in shadcn, the fast path fetches two or three focused URLs.

General library

For broad requests like Next.js docs, expect more URLs and run the analyzer to decide distribution.

Repository fallback

When context7 coverage is weak, use the repository-analysis workflow to inspect source documentation directly.

Sample Prompt

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

Topic lookup Recommended 
/ak:docs-seeker "How do I use date picker in shadcn?"
Use when:
You need current docs for one feature, component, or API pattern.
Expected:
The skill classifies the query as topic-specific, fetches the Context7 topic index with automatic general fallback, reads the few relevant pages, and cites the docs that support the answer.
General library coverage 
/ak:docs-seeker "Documentation for Astro"
Use when:
You need broad library or framework coverage before choosing which pages to read.
Expected:
The skill detects a general query, fetches the library llms.txt index, runs the analyzer to group URLs and suggest distribution, then synthesizes installation, concepts, APIs, and examples from primary pages.
Repository fallback 
/ak:docs-seeker "Find current docs for github.com/pacocoursey/cmdk. If Context7 has no llms.txt, use repository analysis; cloning to /tmp/docs-analysis is allowed."
Use when:
Context7 or official llms.txt coverage is missing and repository evidence is allowed.
Expected:
The skill follows the repository-analysis fallback: find and verify the official repo, optionally clone and pack it with Repomix, extract README/docs/examples/API evidence, and label code-based claims as fallback evidence.

Handled Scope

  • llms.txt and context7 documentation lookup
  • Topic-specific versus general library search
  • Repository documentation fallback analysis
  • URL categorization and parallel research planning
  • Environment-aware script execution with .env fallback order

Artifacts Produced

Documentation Evidence Packet

Pattern

Library/topic classification + fetched URLs + analysis strategy

Includes detected topic or general query, the Context7 source URL, primary docs read from the returned index, fallback path, URL priority, and recommended agent distribution when broad.

Next