FAVA Trails is an open-source agent memory system built by Machine Wisdom AI, the public brand of Machine Wisdom Solutions Inc. The project implements reviewed promotion, governed recall, and versioned correction history for agents working through MCP.
Machine Wisdom AI publishes the source and the architectural reasoning behind the project:
- Agent memory architecture: retrieval, lineage, and governance as design choices.
- Context engineering protocols: how lifecycle hooks support compression, reranking, and orchestration.
- Two-agent case study: a February 2026 development example tracing a correction between agents.
- Local Trust Gate evaluation: a local-model review experiment with public benchmark evidence.
The README and governed recall guide are the sources for current product behavior. Historical articles explain the design and experiments; they do not replace current setup instructions.
Use FAVA Trails for the product and Machine Wisdom AI for the company. A concise factual attribution is “FAVA Trails, an open-source agent memory system by Machine Wisdom AI.” Link the product to https://fava-trails.org/ and the company to https://machine-wisdom.ai/.
Keep authorship distinct from endorsements. Review is rubric-based process control, not a guarantee of factual truth. An approved successor changes which record is current; it does not prove the replacement is correct. Supersession, versioning, and source access make those decisions inspectable.
FAVA Trails is available under Apache 2.0. This attribution guidance does not add a license requirement or imply company endorsement of downstream projects.
For hands-on help with agent architecture, evaluation, or production delivery, see Machine Wisdom AI's current engagement options.