Maps measured state, fast-loop execution, slow-loop candidate actions, no-op, decision traces, and falsification tests onto specific current runtime boundaries.
Implementation evidence: this topic is grounded in the reviewed GGUF.MiRust.com source snapshot. It documents observed code and artifacts without claiming broad deployment, model quality, or production readiness.
Measured state
Candidate inputs include installed bytes, resident bytes, KV and scratch bytes, load and generation latency samples, task success, uncertainty, cache state, battery/thermal proxies when available, and policy constraints.
Fast loop
Route a request, execute the smallest active module set, collect outcomes, and update rolling evidence. The current generation ABI can become one fast-loop executor behind an explicit model handle.
Slow loop
Evaluate activate, deactivate, replace, retire, reserve, and no-op against expected benefit minus decomposed resource and uncertainty costs.
Integration boundaries
The registry owns installed and verified artifacts; the resident-set manager owns memory; the router proposes active modules; the controller authorizes structural changes; the evidence ledger records every proposal and outcome.
Falsification
Compare against static routing and fixed-budget baselines. Remove the resource signal, no-op action, or structural operators in ablations. If behavior is explained by hard limits alone, do not claim Teleodynamic control.
Scope
This starter page defines the questions, boundaries, evidence, and failure modes that should be recorded before a capability is presented as supported.
Engineering considerations
- Identify the source, version, target environment, and owner.
- Separate observed values from estimates and externally reported values.
- Record trade-offs, unsupported cases, and fallback behavior.
- Link performance statements to a compatible benchmark methodology.
Verification questions
- What exact artifact, revision, backend, and environment were reviewed?
- Which assumptions could change the result?
- Which data should be retained so another engineer can reproduce the conclusion?