Skip to content

Memory

Personal memory subsystem — semantic store, mental models, retention, reflection

Details

Persistent personal memory layer for building a digital twin. Captures soft personal context (preferences, habits, emotional state, recurring patterns) that artifacts systematically miss.

Backed by Hindsight, the external server that handles storage, semantic search, and LLM-powered reflection. See memory/hindsight for server components, env vars, and startup details.

Integration paths

Two paths share one local Hindsight server:

  1. Claude Code plugin (hooks) — ambient auto-recall before every prompt, auto-retain after responses.
  2. Python adapter (work_buddy/memory/) — programmatic retain/recall/reflect for context collection, workflows, and MCP gateway capabilities.

MCP capabilities

  • memory_read — semantic + keyword search over stored memories
  • memory_write — store a fact, preference, or constraint
  • memory_reflect — LLM-powered reasoning over memories (consent-gated)
  • memory_prune — delete memories (consent-gated, irreversible)

Bank and tags

One personal bank (user by default, configurable via hindsight.bank_id in config.yaml). Tag taxonomy: user:default, source:*, kind:*, domain:*, workflow:*, session:*. Observations (automatic pattern consolidation) and mental models (pre-computed reflect summaries) build up the digital twin over time.

Bank bootstrap

Call ensure_bank() from work_buddy.memory.setup to create the bank with missions, directives, and mental models. Idempotent — safe to call repeatedly.

Children