Skip to content

MCP Interface

Protocol: Model Context Protocol (MCP)

Bukti exposes six MCP tools, allowing any tool-enabled LLM agent to search for talent, verify capability claims, retrieve provenance chains, and register AI agents.

Tool inventory

Search for entities (persons, AI agents, or both) by natural-language query with optional structured filters such as minimum tier, domain, location, observation date, and agent-specific operational filters (cost, latency, MCP availability, modality).

bukti_verify

Verify a single capability claim for a specific entity. Returns the aggregated confidence and tier together with summary identity and substantive grades.

bukti_capability_provenance

Retrieve the VOI-level evidence detail for a single entity-capability pair. Lazy-loads the evidence chain that is omitted from bukti_search responses to avoid bloat.

bukti_profile

Retrieve a full profile in one of four formats: JSON, machine-readable skills.md for LLM context injection, operational agent.md for AI agent entities, or Schema.org JSON-LD.

bukti_context

Retrieve a profile together with an explicit relevance explanation against a provided query. Designed for agents doing candidate-fit analysis.

bukti_register

Register an AI agent entity in the Bukti capability registry. Accepts the agent's runtime, model identifiers, declared capabilities with supporting evidence, and optional operational metadata (MCP endpoint, cost per call, latency). Returns a job identifier and, on completion, the assigned entity identifier.