35 lines
884 B
Markdown
35 lines
884 B
Markdown
# pipeline_setup_v3
|
|
|
|
`pipeline_setup_v3` is a test harness over agent runtime components.
|
|
|
|
It keeps the same high-level workflow as the previous pipeline setups:
|
|
|
|
- declare YAML test cases;
|
|
- resolve or build `rag_session_id`;
|
|
- run `router_only`, `router_rag`, or `full_chain`;
|
|
- write per-case JSON/Markdown details;
|
|
- write a Markdown summary.
|
|
|
|
The important difference is that `v3` does not assemble a local test-only pipeline.
|
|
It uses agent components directly:
|
|
|
|
- `IntentRouterV2`
|
|
- `RuntimeRetrievalAdapter`
|
|
- `AgentRuntimeExecutor`
|
|
|
|
## Run
|
|
|
|
```bash
|
|
PYTHONPATH=. python -m tests.pipeline_setup_v3.run \
|
|
--cases-dir tests/pipeline_setup_v2/cases/suite_02_pipeline/full_chain_smoke.yaml \
|
|
--run-name smoke_v3
|
|
```
|
|
|
|
## Reindex a repo into RAG
|
|
|
|
```bash
|
|
PYTHONPATH=. python -m tests.pipeline_setup_v3.cli.index_repo \
|
|
--repo-path /abs/path/to/repo \
|
|
--project-id my_repo
|
|
```
|