Hindsight: AI Long-Term Memory
Self-host Hindsight with Pigsty v4.3 and store long-term memory in external PostgreSQL.
Hindsight is a PostgreSQL-native long-term memory service for AI agents.
Pigsty v4.3 provides the app/hindsight configuration template (conf/app/hindsight.yml), using Pigsty-managed PostgreSQL by default instead of Hindsight’s built-in development database.
Quick Start
curl -fsSL https://repo.pigsty.io/get | bash; cd ~/pigsty
./bootstrap
./configure -c app/hindsight
vi pigsty.yml # edit domain, database password, and LLM settings
./deploy.yml
./docker.yml
./app.yml
Default access URLs:
- UI:
http://hs.pigstyorhttp://<IP>:9999 - API:
http://hs-api.pigstyorhttp://<IP>:8888
Key Settings
conf/app/hindsight.yml overrides /opt/hindsight/.env through apps.hindsight.conf. Key parameters:
HINDSIGHT_API_PUBLISH_PORT: public API port, default8888HINDSIGHT_UI_PUBLISH_PORT: public UI port, default9999HINDSIGHT_DB_HOST/HINDSIGHT_DB_PORT/HINDSIGHT_DB_NAME/HINDSIGHT_DB_USER/HINDSIGHT_DB_PASSWORDHINDSIGHT_API_VECTOR_EXTENSION: defaultpgvectorHINDSIGHT_API_TEXT_SEARCH_EXTENSION: defaultnativeHINDSIGHT_API_LLM_PROVIDER: defaultnone
The default none LLM provider only ensures the service can start. Fact extraction, reflection, and consolidation require configuring Ollama or an OpenAI-compatible API.
Operations
cd /opt/hindsight
make up
make log
make info
make down
make pull
References
- Hindsight project: https://github.com/vectorize-io/hindsight
- Pigsty template: https://github.com/pgsty/pigsty/blob/main/conf/app/hindsight.yml
Feedback
Was this page helpful?
Thanks for the feedback! Please let us know how we can improve.
Sorry to hear that. Please let us know how we can improve.