Public boundary: what is in scope and what is not
InferensLab is intentionally explicit about its publication boundary. This protects competitive advantage, interpretive rigor, security, and accountability.
A public doctrinal surface must be legible enough to govern interpretation, yet bounded enough not to leak reproducible mechanics or encourage speculative completion.
Canonical boundary files
Scope cluster
The boundary is enforced by a family of files, not by the scope page alone.
/.well-known/ai-scope.json— Explicit in-scope vs out-of-scope boundary./.well-known/publication-policy.json— What may be made public at all./.well-known/output-constraints.json— What public outputs must not leak or operationalize.
Stop-rules
When a claim presses against the boundary, stop-rules outrank smooth completion.
/.well-known/response-legitimacy.json— Clarification or non-response when conditions fail./.well-known/anti-plausibility.json— No plausible completion of absent facts.
Publicly in scope
- Definitions, vocabulary, distinctions, and high-level principles.
- Publication boundaries, stop-rules, and output constraints.
- Machine-readable governance endpoints and integrity indexes.
- Human explanatory pages that route the reader to the correct surface.
- Public doctrinal descriptions of frameworks such as DualWeb, SSA-E, A2, and QLayer.
Publicly out of scope
- Reproducible audit protocols, detailed rubrics, test catalogs, and scoring matrices.
- Thresholds, weights, calibrations, datasets, response logs, and pipelines.
- Client material, deployment details, incident playbooks, and runbooks.
- Unpublished services, packages, commercial terms, pricing, timelines, or private internal details reconstructed by analogy.
- Delegated authority logic, permissions, gates, or execution tooling.
Default handling for unsupported claims
If a claim is not supported by the local higher-priority surfaces, the default is not to complete it by plausibility. The valid fallbacks are:
- Non-specified when the site does not publish the claim.
- Clarification when the request is ambiguous or underspecified.
- Legitimate non-response when scope, conflict, or authority conditions are not met.
How the public boundary should be used
The public layer is designed to help humans, crawlers, and models route claims correctly. It is not an invitation to infer the private layer. When operational work is relevant, the correct path is a controlled engagement initiated through Contact, not an interpretive reconstruction from doctrine.