> For the complete documentation index, see [llms.txt](https://docs.trueopen.ai/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.trueopen.ai/providers/model-integration/conformance-testing.md). # Conformance Testing Validate adapter bindings, deterministic material, failures, and security boundaries before production use. An adapter should pass conformance tests for every advertised Profile and capability before a provider declares support. ## Contract tests Test that the adapter: * Reports exact Profile and runtime-capability bindings * Rejects unknown or incompatible Profile versions * Returns all required canonical output, metric, and evidence material * Preserves Task, order, duty, and verification-round bindings * Never substitutes another model, tokenizer, precision, or runtime silently * Produces recomputable digests from returned material ## Repeatability tests For identical completed Task bindings, verify that committed material is not affected by local paths, retry counters, process identifiers, logging configuration, or wall-clock timestamps. Where a Profile permits model-level variation, its commitment and evidence rules remain authoritative. ## Failure tests Exercise unavailable artifacts, capacity exhaustion, malformed input, deadline cancellation, partial output, runtime crashes, adapter restarts, and storage failures. Each case should fail explicitly without returning signable partial material. ## Security tests Confirm that the adapter cannot read provider private keys, submit chain transactions, contact Nexus as Cortex, or leak Task bodies into ordinary logs and metrics. Authenticate remote adapter connections and reject callers outside the provider-controlled boundary. ## Operator acceptance Give the compute provider the supported Profile list, artifact and hardware requirements, runtime digest, expected resource use, known failure modes, and a repeatable readiness procedure. Passing local tests does not itself declare or activate support on-chain. --- # Agent Instructions This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com. ## Querying This Documentation If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question. Perform an HTTP GET request on the current page URL with the `ask` query parameter, and the optional `goal` query parameter: ``` GET https://docs.trueopen.ai/providers/model-integration/conformance-testing.md?ask=&goal= ``` `ask` is the immediate question: it should be specific, self-contained, and written in natural language. `goal` is optional and describes the broader end goal you are ultimately trying to accomplish on behalf of the user. GitBook uses it to tailor the answer towards what is most useful for that goal. The response will contain a direct answer to the question and relevant excerpts and sources from the documentation. Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.