Four-Leaf MCP - Interview prep + job search inside the AI you already use
by•
Four-Leaf is now a Model Context Protocol server. Install it in Claude, Cursor, ChatGPT, Cline, Continue, Windsurf, or Perplexity, and your AI assistant gets 11 job-search and interview-prep tools: search 100k+ active job postings, score your resume against a JD, generate practice questions, get cited salary bands, and run a full negotiation strategy.
Two paid surfaces deep-link to four-leaf.ai for voice mock interviews and AI resume tailoring. Open source MIT.
Replies
Hey! I built this because every AI interview prep tool today is either trapped in its own app or producing generic hallucination. Once your AI assistant has real tools for job search, practice questions, role intel, and comp analysis, prep gets faster and grounded.
The MCP itself is hosted. The Skill wrapper that calls it is MIT at github.com/fourleafai/clover-public.
Architecture write-up if you're curious: four-leaf.ai/blog/job-search-assistant-mcp. OAuth 2.1 + PKCE + DCR, server-side web search for grounded comp data, single-use stash tables for heavy text handoff.
Happy to answer questions about the build.
Mailwarm
What sources are you using for the salary data, and does it adapt by city and seniority?
@naimz The tool runs a real web search server-side, then synthesizes the band from what it finds. The model is steered toward known-good sources: levels.fyi, Glassdoor, salary.com, Built In, Payscale, public H1B disclosure data, and sector salary surveys for niche roles. Every claim in the response cites the source by name with a confidence rating.
Inputs are role, level, location, and optional company, so yes, it adapts by city and seniority. A senior backend engineer in Austin gets a different band than the same role in NYC.