iuTdZddlZddlmZddlmZddlmZddlmZm Z de d e fd Z y) u5 Genesis Persistent Context Architecture — JIT Hydration Master Function Story 2.02 — Track B interceptor_jit_hydration() — Master JIT Hydration Function. Assembles envelope from all memory layers concurrently. Every LLM call starts with full context in <50ms. Pipeline: 1. fast_extract(task_payload) -> (target_entities, intent_string) 2. scatter_gather_memory(...) -> MemoryContext [3 layers, 45ms timeout] 3. build_envelope(memory_context) -> XML string 4. Attach as task_payload["system_injection"] N)Optional) fast_extract)scatter_gather_memory)build_envelope MemoryContext task_payloadreturncKt|\}}|jdd}t|||dd{}t|}||d<|S7w)u Master JIT hydration function. Orchestrates the full memory assembly pipeline for a single LLM call: Step 1 — fast_extract() (synchronous, pure Python, <1ms) Parses the task payload and extracts entity names and an intent string. Never makes I/O calls. Safe to run in the hot path. Step 2 — scatter_gather_memory() (async, 45ms wall-clock budget) Fires L1 (Redis), L2 (KG JSONL), and L3 (Qdrant) fetches concurrently. Each layer has its own timeout. A slow or offline layer returns None without blocking the others. MemoryContext always returns with whatever data arrived in time — graceful degradation is built into scatter_gather. Step 3 — build_envelope() (synchronous, pure Python, <1ms) Assembles the MemoryContext into a well-formed XML string. None fields are replaced with canonical fallback strings so the envelope is always non-empty and valid. Step 4 — Attach envelope to task_payload["system_injection"] The original payload is mutated in-place and returned. All pre-existing keys are preserved. Args: task_payload: Dict representing the LLM task. May contain any of: - "task_id" : str — used as Redis L1 lookup key - "prompt" : str — used for entity extraction + intent - "description" : str — fallback text source for entity extraction - "task" : str — fallback text source for entity extraction - "file" : str — text source for entity extraction - "code" : str — text source for entity extraction - "context" : str — text source for entity extraction Returns: The same task_payload dict with one additional key: task_payload["system_injection"] — str, the XML envelope task_idunknown-)r target_entities intent_string timeout_msNsystem_injection)rgetrr)r rrr memory_contextenvelopes 2/mnt/e/genesis-system/core/memory/jit_hydration.pyinterceptor_jit_hydrationrskP&2,%?"O]y)4G*?'# +%N#>2H(0L#$ %s3AA A) __doc__asynciotypingrrscatter_gatherrzero_amnesia_enveloperrdictrrr s. &1@9$949r