Kartikeya
2223b190f6
infra: Dockerfile (backend from monorepo root) + DigitalOcean App Platform spec with health check and secret env vars. database: MongoDB Atlas + Voyage vector-memory schema and the continual-learning loop notes. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
1.6 KiB
1.6 KiB
database
Continual-learning memory for PodMan: MongoDB Atlas for the team model and outcomes, Voyage embeddings for vector recall. This is what makes PodMan "more useful the more you use it" (the track requirement).
Collections
| Collection | Holds | Notes |
|---|---|---|
pods |
Pod docs (members, repo) |
one per pod |
observations |
EngineerContext snapshots over time |
sampled, append-only |
collisions |
detected Collisions |
for replay + precision tuning |
interventions |
Interventions + outcome (accepted/dismissed) |
drives the self-tuning policy |
memory_vectors |
Voyage embeddings of file/feature notes | Atlas Vector Search index |
Types live in shared/. Each engineer/file/feature note is
embedded with Voyage and stored alongside its source doc for retrieval by the
PodMan brain.
The continual-learning loop
- Observe → write
observations. - Store → embed notes into
memory_vectors. - Predict → collision detector + brain decide whether to intervene.
- Outcome → update the
interventionsdoc with accepted/dismissed. - Adapt → tune thresholds + ownership attribution from outcomes.
Setup
Create an Atlas cluster + a Vector Search index on memory_vectors.embedding,
then set MONGODB_URI and VOYAGE_API_KEY in .env.