pipeline_ops/skill/SKILL.md

30 lines
1018 B
Markdown

---
name: pipeline-ops
description: "Pipeline operations module — pricing, capacity management, and usage logging."
---
# pipeline_ops
Operations layer: pricing configuration, capacity quotas, and usage logging for pipeline billing.
## Data Model
| Table | Key Fields |
|-------|-----------|
| `pipeline_pricing` | id, pipeline_id, model_name, unit_price, price_unit |
| `pipeline_capacity` | id, pipeline_id, max_concurrent, max_daily |
| `pipeline_usage_log` | id, pipeline_id, distributor_id, tokens, cost |
## DSPY Endpoints
- `pipeline_pricing_create/update/delete.dspy` — pricing CRUD
- `pipeline_capacity_create/update/delete.dspy` — capacity CRUD
- `pipeline_usage_log_create/update/delete.dspy` — usage log CRUD
- `ops_dashboard_data.dspy` — aggregated stats dashboard
- `get_search_pipeline_id.dspy` — pipeline dropdown search
## Pitfalls
- Usage log writes must be atomic — use `sor.C` inline, not batch.
- `ops_dashboard_data.dspy` aggregation queries — test with realistic volume.