Every analytics tool wants a subscription. And your data. I built Twiligent instead. YouTube stats. Instagram stats. One dashboard. Runs locally on your machine.
MCP: Parallel Search - Research
Purpose
Parallel Search fetches current web content for research tasks: UX patterns, browser compatibility, best practices, competing implementations, and up-to-date API behavior when Context7 doesn't have it.
When to Use
Use Parallel Search for:
- UX patterns for a feature you're designing ("how do social media schedulers handle bulk post editing?")
- Browser/platform compatibility ("does Instagram Graph API support carousel posts on all account types?")
- Recent API changes not yet in Context7 ("Instagram Graph API v21 vs v25 container creation changes")
- Accessibility guidelines ("WCAG criteria for data table accessibility")
- Competitor analysis ("how does Buffer/Hootsuite handle the scheduling queue UI?")
- Current events ("is Upstash Redis still recommended for Cloudflare Workers in 2025?")
Do not use Parallel Search for:
- Library API docs with stable method signatures (use Context7)
- Codebase exploration (use Filesystem)
- Architecture planning (use Sequential Thinking)
- When you already know the answer from existing docs
Research Workflow
UX Research for a New Feature
- Identify what you're designing (e.g., a post scheduling calendar view)
- Use Parallel Search with 2-3 queries:
- "social media scheduling calendar UX patterns 2024"
- "how Hootsuite implements scheduling calendar"
- "time slot selection UI accessibility"
- Extract the patterns relevant to this project's constraints (mobile-first, React)
- Apply the patterns - document sources in a code comment if the pattern is non-obvious
API Behavior Research
When Instagram/YouTube API behavior is unclear or seems different from Context7 docs:
- Search for recent issues or changelogs
- Check the official API changelog (graph.facebook.com/changelog, developers.google.com/youtube/v3/revision_history)
- Look for Stack Overflow or GitHub issues with the specific error/behavior
Accessibility Research
Before building any interactive feature:
- Search "WCAG 2.1 {feature type}" - e.g., "WCAG 2.1 date picker accessibility"
- Check MDN for the correct ARIA roles and attributes
- Apply what you find - shadcn components already include correct ARIA in most cases
Multiple Search Queries
Parallel Search can issue multiple queries at once. When researching a feature, batch related queries:
Search queries:
1. "instagram graph api reels container creation required fields 2025"
2. "instagram graph api video processing status polling best practices"
3. "instagram reels api error codes"
This returns results faster than sequential searches.
Integration with Implementation
Research findings should inform the implementation but not be pasted verbatim into code. Extract:
- The pattern or API contract
- Any constraints or gotchas
- The recommended approach
Then implement using the project's own style and conventions.