LYNK is the EXOCHAIN adapter package for services that need to withhold model or tool output until an AVC-backed usage receipt is accepted by EXOCHAIN core.
The adapter wraps OpenAI Responses, OpenAI Chat Completions, and MCP tools/call usage in signed AVC evidence, then submits it to the core receipt endpoint.
EXOCHAIN validates the AVC evidence and emits the node-signed receipt through POST /api/v1/avc/llm-usage/receipts/emit.
This public page is an adjacent public surface. It does not mint receipts, validate credentials, or imply constitutional enforcement beyond the tested EXOCHAIN core/API receipt path.
LYNK receipts are designed to carry hashes, integer usage counters, safe metadata, custody policy hashes, and receipt/finality links. They are not a place for prompts, completions, tool arguments, tool results, provider keys, bearer credentials, KMS material, object-store locations, or decryptable payload material.
name: EXOCHAIN LYNK Protocol classification: core runtime adapter public_site_classification: adjacent public surface v1_positive_lanes: openai_responses, openai_chat_completions, mcp_tools_call receipt_endpoint: POST /api/v1/avc/llm-usage/receipts/emit default_custody_mode: receipt_minimized unsupported_claims: site_enforcement, release_readiness_without_gates secret_material_in_public_copy: forbidden
Provider usage is summarized into signed AVC evidence before receipt emission.
Output delivery remains gated on the receipt outcome reported by EXOCHAIN.
Tool usage can produce signed LYNK evidence; EXOCHAIN core emits the receipt.
Unsupported until a separate provider adapter and positive tests land.
Unsupported beyond the already tested OpenAI-compatible v1 paths.
May create signed evidence; must not mint EXOCHAIN receipts directly.
For broader EXOCHAIN concepts, read the AVC explainer, trust receipt anatomy, and developer guide.