Bryan Helmkamp 92a4c5e506 Inject fidelity preamble into LLM prompts and reuse sessions for full fidelity ... CodergenHandler now reads `current.preamble` from context and prepends it to the prompt before dispatching to backends. For `full` fidelity the engine sets preamble to empty, so the prepend is a no-op. This ensures downstream LLM nodes see prior node outputs (e.g. script stdout) without any trait changes. AgentBackend gains a `sessions` cache keyed by thread ID. When fidelity is `full` and a thread_id is present, sessions are reused across nodes so the LLM sees the full conversation history. Usage aggregation only counts turns added after the reuse point to prevent double-counting. Failed sessions are dropped rather than cached. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>		2026-02-27 12:27:36 -05:00
..
agent	Fix pre-existing lint warnings	2026-02-27 12:09:48 -05:00
attractor	Inject fidelity preamble into LLM prompts and reuse sessions for full fidelity	2026-02-27 12:27:36 -05:00
git-storage	Add git-storage crate for storing data in git object database	2026-02-24 23:17:44 -05:00
llm	Add Claude Haiku 4.5 to model catalog	2026-02-25 14:21:58 -05:00
redact	Add redact crate for secret detection and redaction in NDJSON logs	2026-02-25 15:25:52 -05:00
terminal	Revert clippy config to defaults, remove all pedantic/nursery/cargo lint suppressions	2026-02-24 09:42:28 -05:00