honcho-aio dub19/honcho-aio https://hub.docker.com/r/dub19/honcho-aio bridge sh false https://github.com/wgross19/honcho-aio/issues https://github.com/wgross19/honcho-aio Honcho is memory infrastructure for agents that track people, sessions, and ideas over time. The default setup runs Honcho with its internal Postgres and Redis services. Store messages. Let it derive. Query context or search. honcho-aio packages that for Unraid. One container. Honcho on port 8000. Postgres and Redis inside, not on the LAN. Set a password and <provider>/<model> strings. Then hermes memory setup → http://<lan>:8000. First run: 1. Set PostgreSQL Password (required). The container exits 64 if it is empty. 2. Set EMBEDDING_MODEL and CHAT_MODEL as <provider>/<model>. Split is on the first /. Nested model slashes stay (openrouter/openai/gpt-5.4-mini). 3. If you use local/, set LOCAL_BASE_URL (example http://192.168.1.79:11434/v1). Local chat has no cloud fallback. 4. Fill only the API key for the provider you picked. 5. Start, then point Hermes with: hermes memory setup → http://<lan>:8000 The chat model must support tool/function calling. Advanced configuration can select a local or cloud provider and custom model settings. At boot the image probes embedding width. A mismatch with EMBEDDING_VECTOR_DIMENSIONS fails boot. ### 2026-08-23 - Generated from CHANGELOG.md during release preparation. Do not edit manually. - Complete honcho-aio Dockerfile with s6-overlay, PG17+pgvector, Redis, Honcho v3.0.12 compile - Fix s6 services + cont-init bootstrap to match spec - Provider abstraction — <provider>/<model> parsing + dimension guard - Unraid XML template (honcho-aio.xml) - Rewrite honcho-aio.xml to Unraid CA Config attribute shape - Track generated .aio-fleet.yml fleet contract - Validate + pytest (10 unit, 9 integration) + trunk green - Reconcile pgvector dims with EMBEDDING_VECTOR_DIMENSIONS at api start - Catalog-ready XML and published fleet contract - Set fleet upstream version_key to HONCHO_VERSION AI Tools:Utilities http://[IP]:[PORT:8000] https://raw.githubusercontent.com/wgross19/awesome-unraid/main/honcho-aio.xml https://raw.githubusercontent.com/wgross19/awesome-unraid/main/icons/honcho.png honcho memory agent postgres pgvector redis hermes Docker. A LAN IP you control. Optional: a local OpenAI-compatible embed/chat endpoint for local/ models. Support wgross19 on GitHub Sponsors. https://github.com/sponsors/wgross19 Honcho is memory for stateful agents. It stores messages, reasons in the background, then returns context, search, and peer conclusions. honcho-aio runs that stack as one image: Honcho API, PostgreSQL 17 + pgvector, Redis, and s6. Databases stay on loopback. Point Hermes at http://<lan>:8000. No MCP. bridge 8000 8000 tcp /mnt/user/appdata/honcho-aio/postgres /data/postgres rw /mnt/user/appdata/honcho-aio/redis /data/redis rw /mnt/user/appdata/honcho-aio/honcho /var/lib/honcho rw 8000 /mnt/user/appdata/honcho-aio/postgres /mnt/user/appdata/honcho-aio/redis /mnt/user/appdata/honcho-aio/honcho postgres honcho local/embeddinggemma openai/gpt-5.4-mini 768 ollama false false false false 4 WARNING default