project-brain

To maintain AI agent memory across sessions, load a markdown brain document at the start to restore context and save it at the end using incremental merges. This persists agent state and decisions between runs without losing continuity.

Session memory loading auto-discovers a brain document, reads structured sections, and briefs the current state. Saving creates or updates the document using incremental merges, preserving history while applying curated safety guards that require user confirmation before writes.

Supported formats for persisting AI session context include v2.0, v1.0, and free-form markdown brain documents. The system auto-discovers these structured files to read sections and brief the current state before applying incremental updates.

Yes, you can use this session memory tool with Obsidian vaults, sprint-forge, and universal-planner. It integrates directly with these platforms to load and persist agent state, ensuring workflows maintain continuity across different task environments.

No, the agent memory system does not apply silent defaults. It uses curated safety guards that require explicit user confirmation before writing updates to the brain document, ensuring you maintain full control over state migrations and incremental merges.

AI agents lose context between different task runs because they lack persistent memory to recover state. Solving this requires loading a markdown brain document at the start of a session to restore continuity and saving decisions at the end for future recovery.