NIGHTCRAWLER: Temporal Causal Closure for Persistent and Deferred Effects of Autonomous AI Agents

Autonomous software agents can leave delayed messages, workflows, delegated credentials, and other residual mechanisms after their initiating processes stop. NIGHTCRAWLER develops a conditional formal framework for investigating this residual effect surface in initially unmediated environments. The work distinguishes historical causation from future activation, builds an evidence-conditioned authority and provider scope, analyzes horizon-bounded residual execution paths, proposes invariant-constrained neutralization, and defines four scope-relative attestation outcomes. The results are conditional on explicit coverage and provider-evidence contracts. Executable tests are model-relative consistency checks; no real-provider effectiveness is asserted. This is a research preprint, not a deployed security product.