context-memory

To save and load session context in a chained prompt, use dedicated lifecycle commands like SAVE, LOAD, and INJECT. This ensures analysis results and session data stay coherent and are carried over automatically during long multi-skill workflows.

Multi-skill workflows lose context between chained prompts because session data is not persisted by default. Implementing a defined memory schema to write to disk solves this by allowing subsequent skill calls to recall previous analysis results and context.

The best way to carry over trade setups and analysis results is using a persistent in-memory to disk storage flow. This approach remembers the last analyzed pair and current regime to tailor the next steps in your trading assistant workflow.

Yes, saving session context to disk works for long trading assistant workflows by persisting a memory store to a local JSON file. This enables context carryover, ensuring conversations stay coherent across multiple steps and future interactions.

You should clear persistent memory in a multi-skill pipeline when switching to a completely new analysis topic or ending a long session. Using a dedicated CLEAR command removes outdated trade setups and context, preventing stale data from affecting future skill invocations.