mcp-based people search integration
Exposes Pearch's people search engine as an MCP (Model Context Protocol) server, allowing Claude and other MCP-compatible AI agents to query talent databases through standardized tool-calling interfaces. Implements MCP resource and tool schemas to abstract away HTTP API complexity, enabling agents to discover and filter people by skills, location, experience, and other professional attributes without direct API management.
Unique: Wraps a specialized people search engine (Pearch) as a standardized MCP tool, eliminating the need for agents to manage HTTP authentication, pagination, or API versioning — agents interact via declarative tool schemas instead
vs alternatives: Simpler than building custom Claude plugins or function-calling wrappers because MCP handles protocol negotiation and tool discovery automatically; more specialized than generic web search because it indexes professional profiles and skills
talent attribute filtering and search
Provides structured search capabilities to filter candidates by professional attributes including skills, geographic location, years of experience, job titles, and employment status. Implements query translation from natural language (via Claude) into Pearch's backend search API, supporting multi-field filtering and ranking by relevance. Abstracts backend search syntax so agents can express intent declaratively without learning Pearch's query language.
Unique: Specializes in professional attribute filtering (skills, experience, location) rather than generic full-text search; leverages Pearch's curated people index which is pre-processed for professional context (job titles, skill extraction, employment status)
vs alternatives: More precise than LinkedIn's public search API because Pearch indexes structured professional data; faster than manual recruiter outreach because filtering happens server-side with pre-indexed attributes
agent-driven candidate discovery workflow
Enables multi-step agentic workflows where Claude or other MCP clients iteratively refine candidate searches, evaluate results, and trigger follow-up actions (e.g., outreach, profile deep-dives). Implements tool composition patterns where search results feed into downstream tools, allowing agents to autonomously discover candidates, assess fit, and prepare recruitment actions without human intervention between steps.
Unique: Leverages MCP's tool composition model to enable agents to chain search, evaluation, and action steps without explicit orchestration code — agents autonomously decide when to refine searches or trigger outreach based on intermediate results
vs alternatives: More flexible than rigid recruitment pipelines because agents can adapt strategy based on results; more autonomous than manual sourcing because it eliminates human decision points between search and outreach
natural language to structured search translation
Translates free-form natural language queries (e.g., 'Find senior backend engineers in NYC who know Rust') into structured search parameters (skills array, location, experience level) that Pearch's backend can execute. Leverages Claude's language understanding to parse intent, extract entities, and map them to Pearch's searchable attributes. Handles ambiguity resolution (e.g., 'NYC' → location filter) and skill name normalization without requiring users to learn Pearch's query syntax.
Unique: Bridges conversational intent and structured search by using Claude to parse natural language into Pearch's filter schema — eliminates the need for users to understand backend query syntax while maintaining precision through structured output
vs alternatives: More user-friendly than direct API calls because it accepts natural language; more accurate than simple keyword matching because it leverages LLM entity extraction and semantic understanding
candidate profile enrichment and context injection
Retrieves and enriches candidate profiles with additional context (employment history, portfolio links, social profiles) from Pearch's database, then injects this context into Claude's conversation for deeper analysis. Enables agents to make informed decisions about candidate fit by providing comprehensive professional background without requiring separate API calls or manual profile lookups. Implements context windowing to balance information richness with token efficiency.
Unique: Integrates profile enrichment directly into the MCP tool layer, allowing agents to access comprehensive candidate context without separate API calls or manual lookups — profiles are pre-fetched and injected into Claude's reasoning context
vs alternatives: More efficient than manual profile review because enrichment is automated; more contextual than search-only workflows because agents have full professional background for decision-making