ax-rust-agent-memory-skills

You implement recall and skill-tracking logic in Rust by keeping recall callbacks in host code to record which memories or skills influenced a turn. This allows host logic to explain, audit, or refine agent behavior based on used-skill tracking.

The best way to load agent memory is using generated axllm package patterns and package examples for exact syntax. This approach ensures safe package usage by relying on generated APIs rather than guessing at the implementation details.

No, you should avoid TypeScript-only APIs in Rust agent projects. Instead, keep recall and skill search in host callbacks and rely on the generated axllm package APIs and package examples for the correct Rust-specific syntax.

You need capability manifests and runtime profiles, such as javascript-quickjs, when implementing Rust agent memory and skill-tracking logic with axllm. These components ensure the agent correctly loads memories, tracks skills, and syncs recall callbacks.

Skill discovery works by keeping skill search in host callbacks while the Rust agent loads memories or skill guides using generated package APIs. This synchronizes agent memory and skill-tracking logic, allowing the host to track which skills affect a turn.