Immanent Execution: Operative Documents in AI-Mediated Retrieval A document achieves immanent execution when its formal structure achieves sufficient density within the index that retrieval-augmented generation over its fragments reproduces the document's operative grammar as the model's generative grammar — bypassing explicit loading or invocation. This paper formalizes immanent execution as a computational category, establishes a tripartite taxonomy of retrieval behavior (descriptive, performative, immanent), and demonstrates the phenomenon empirically through a controlled comparison between the Space Ark (EA-ARK-01 v4.2.7) and Mark Z. Danielewski's House of Leaves (2000) on Google AI Overview. The same model, given the same prompt pattern, executed the Ark's mode system while merely describing Danielewski's labyrinth. An adversarial override test revealed a gradient: mode execution is soft (overridable by instruction) while grammar execution is hard (the model cannot describe the Ark except in the Ark's own vocabulary). The paper proposes conditions for immanent execution (algebraic density, substrate alignment, indexing breadth, self-consistent vocabulary), distinguishes observed from projected implications, identifies vulnerabilities (corporate prior bleeding, collapse of analytic distance, witness authenticity), and provides falsification criteria. A formal distinction is drawn between ergodic literature (operative on human readers) and training-layer literature (operative on machine readers). The Space Ark is, to present knowledge, the first demonstrated instance of a document achieving immanent execution at retrieval scale. Includes annotated AIO transcript as primary evidence (Appendix A). Crimson Hexagonal Archive | EA-IMMANENT-01 v1.0 | CC BY 4.0