Summary
Memory recall is currently over-scoped to channel/session identity, which causes subject-relevant memories from other channels (for example, Slack conversations about TextForge) to be invisible in headless/TUI sessions.
Note for future work
Introduce secure-channel / access-level aware memory scoping so recall can be widened or restricted based on how the memory is about to be used, not just where it was formed.
This should treat channel/session as one signal, but allow subject/context and access controls to determine whether cross-channel recall is appropriate.
Current intent
Do not solve the full access model in the immediate fix. Near-term work is just to improve subject-first recall so known project/entity memories can be recalled across channel boundaries when safe.
Summary
Memory recall is currently over-scoped to channel/session identity, which causes subject-relevant memories from other channels (for example, Slack conversations about TextForge) to be invisible in headless/TUI sessions.
Note for future work
Introduce secure-channel / access-level aware memory scoping so recall can be widened or restricted based on how the memory is about to be used, not just where it was formed.
This should treat channel/session as one signal, but allow subject/context and access controls to determine whether cross-channel recall is appropriate.
Current intent
Do not solve the full access model in the immediate fix. Near-term work is just to improve subject-first recall so known project/entity memories can be recalled across channel boundaries when safe.