Module llm_lifecycle 

CLI harness for native GGUF lifecycle: discover → mmap → infer → evict.

Uses the in-process stack (gguf_bridge / wgpu / Phase 8 bifurcated compute), not Ollama or external daemons. Models are memory-mapped via memmap2; the UI RAM ceiling is tracked separately from the 42 MB SlgArena Sentinel budget.

Structs

CliSession 🔒

Statics

CLI_SESSION 🔒

Functions

clear_session 🔒

cli_read_intent 🔒

default_vault_path

Default vault path when none is supplied (Windows-friendly).

init_log_stream

Attach structured logging and optional 100 ms telemetry samples to stdout.

parse_model_id_ref 🔒

run_duplicate_audit

run_eval

run_evict

run_list

run_load

run_status

session_lock 🔒

store_session 🔒

with_session 🔒