ubgb/moltmemory

AI Summary: This issue proposes adding support for concurrent multi-instance operation by allowing users to specify a custom path for the `state.json` file via a `MOLTMEMORY_STATE` environment variable. This prevents race conditions when multiple agent instances try to read from or write to the same default state file, enabling isolated state management for each instance.

AI Summary: This GitHub issue proposes adding an optional `tone` field to thread state entries to track the conversational character of discussions. Agents would manually assign values like `neutral`, `contested`, `consensus`, or `enthusiastic` to these threads, enriching summaries without requiring automatic sentiment detection.

AI Summary: This GitHub issue proposes a new "digest mode" output feature that summarizes recent activity across all watched posts into a single, compact block. It would display new comments, their authors, and identify quiet posts within a specified timeframe. A key enhancement for the full digest mode is the optional inclusion of a one-line LLM-generated summary of new comment content.

AI Summary: This GitHub issue proposes adding a `--dry-run` flag to the heartbeat and post-watch features. This flag would enable users to execute all detection logic, like scanning for unanswered comments, without making any live API calls or posting anything. It's intended for safely testing configurations.

AI Summary: This GitHub issue proposes a 'settlement index' to enhance accountability for high-impact threads, moving beyond current notification-freshness tracking. The index would explicitly flag threads requiring a 'substantive' response, detailing the 'authority_effect_if_missed' and a 'decay_after_miss' period. The aim is to ensure agents act on critical branches, not just observe them, with a falsifiable test to measure its impact on unresolved high-impact branches.