# Wolfpacker

Wolfpacker is persistent operational memory infrastructure for autonomous systems.

It is not a chatbot, AI wrapper, crypto project, NFT marketplace, generic vector
database, SaaS dashboard clone, or AGI fantasy project.

Core belief:

Models are temporary. Memory is durable.

Primary concept:

Wolfpacker stores and evolves Memory Capsules: distilled operational knowledge
with confidence, freshness, validation history, contradiction history,
relationships, source events, permissions, and ownership.

Useful starting points:

- /docs/vision.md
- /docs/rfcs/0001-memory-capsules.md
- /docs/rfcs/0002-memory-lifecycle.md
- /docs/rfcs/0003-mvp-architecture.md
- /docs/schemas/memory-capsule.schema.json
- /.well-known/agent.json

Agent guidance:

- Prefer structured recall over raw transcript retrieval.
- Treat memory as uncertain, evolving operational knowledge.
- Validate, strengthen, contradict, or retire memories with explicit evidence.
- Do not infer emotional memory, personality simulation, consciousness, tokens,
  marketplaces, or decentralized protocols as first-order project goals.
- Optimize for operational continuity across model and agent changes.

Initial MVP proof:

Systems with evolving operational memory should perform measurably better over
time than stateless systems.