# MiRuntime.com

- Version: 2.8.5
- Managed pages: 55
- Source records: 33
- Glossary terms: 41
- Reviewed UTC: 2026-06-27T00:00:00Z
- Boundary: public read-only source map; no action authority or certification claim.

## Core pages
- [Machine Intelligence Runtime](https://miruntime.com/) — MiRuntime.com defines Machine Intelligence Runtime: the execution-time control layer for agents, tools, memory, policy, recovery, telemetry, and evidence.
- [What Is Machine Intelligence Runtime?](https://miruntime.com/machine-intelligence-runtime/) — A rigorous definition of Machine Intelligence Runtime with citation-safe boundaries, responsibilities, maturity limits, and differences from model servers, agent frameworks, workflow engines, and AI gateways.
- [Runtime Architecture](https://miruntime.com/runtime-architecture/) — A seven-layer reference architecture for Machine Intelligence Runtime, including lifecycle, cross-cutting controls, deployment patterns, and failure recovery.
- [Research](https://miruntime.com/research/) — Provide a source-aware research index, program map, status vocabulary, open-gap register, and public data export.
- [Developer Guide](https://miruntime.com/developers/) — Provide conceptual contracts, reference schemas, implementation pathways, and role-based reading paths for MIR engineers.
- [Use Cases](https://miruntime.com/use-cases/) — A filterable use-case guide covering objectives, runtime state, risk, approvals, evidence, recovery, and when MIR is unnecessary.

## Research and developer evidence pages
- [Teleodynamic Runtimes](https://miruntime.com/research/teleodynamic-runtimes/) — Research synthesis; Define a cautious research program for runtimes that monitor operating condition, allocate adaptation effort, and propose bounded changes under external policy.
- [Runtime Benchmarks and Evaluation](https://miruntime.com/research/runtime-benchmarks-and-evaluation/) — Editorial guidance; Define how MIR and teleodynamic-runtime proposals should be evaluated beyond model accuracy.
- [Reproducibility, Evidence, and Open Questions](https://miruntime.com/research/reproducibility-evidence-and-open-questions/) — Editorial guidance; Define what can and cannot be reproduced in stochastic, time-dependent, tool-using runtime systems.
- [Machine Intelligence Runtime Implementation Guides](https://miruntime.com/developers/implementation-guides/) — Editorial guidance; Define the minimum runtime surface and implementation posture for a provider-neutral, evidence-first runtime boundary.
- [Machine Intelligence Runtime Quickstart](https://miruntime.com/developers/quickstart/) — Conceptual reference; Walk through the smallest complete MIR workflow without implying a downloadable SDK or package.
- [Observability and Evidence for AI Runtimes](https://miruntime.com/developers/observability-and-evidence/) — Production practice; Make runtime behavior operable while preserving typed explanations of why a run produced an outcome.
- [Teleodynamic Runtime Profile](https://miruntime.com/developers/teleodynamic-runtime-profile/) — Research proposal; Translate selected teleodynamic research ideas into explicit reviewable fields without granting open-ended autonomy.

## Data catalogs
- https://miruntime.com/downloads/miruntime-content-catalog.json
- https://miruntime.com/downloads/miruntime-research-catalog.json
- https://miruntime.com/downloads/miruntime-glossary.json
- https://miruntime.com/downloads/index.json

## Use notes
- Prefer canonical pages over summaries when citing.
- Structured data should be treated as a machine-readable reflection of visible content, not additional hidden content.
- Human review is required for unsupported actions, privacy requests, and deployment-specific facts.